{"generated":"2026-10-08T10:51:03.900407Z","pages":[{"description":"Put your company's details on your invoices, and lift the 3-seat limit on Team plans.","gate":null,"markdown":"\nIf you're paying for Crunch through a company, you'll usually want the company's name, tax number\nand address on your invoices — that's what most finance teams need to file them. **Business\nDetails** is where you enter them. It's optional, but once your details are approved you also get a\none-time credit and, on a Team plan, room to grow past three seats.\n\n## What you'll need\n\nYour company's information to hand:\n\n- business name;\n- business type;\n- registration number;\n- tax ID / GSTIN (GSTIN is India's GST registration number);\n- address.\n\n## Step by step\n\n1. In the hub, open **Business Details**.\n2. Fill in your business name, type, registration number, tax ID / GSTIN and address.\n3. Click **Save Business Details**.\n\nSaving sends your details for review. It is usually reviewed within a business day.\n\n## How to tell it worked\n\nOnce your details are approved, three things change:\n\n- **They appear on your invoices**, so every invoice you [download from Billing](/help/hub/billing/)\n  carries your company's details.\n- **Your account gets a one-time credit.** Credits pay for usage beyond what your plan includes —\n  see [Credits](/help/hub/credits/).\n- **A Team plan can grow past 3 seats.** A seat is one person on the plan. While your details are\n  awaiting review, an account is limited to 3.\n\n## Good to know\n\n- **Changing your details starts a new review.** If you edit them later — a new address, say —\n  they're sent for review again.\n- **Planning a bigger team?** If you expect more than three people on a Team plan, save your\n  business details early so the review is done by the time you need the extra seats. Plans and\n  seat ranges are on [pricing](/pricing/).\n\n## Related\n\n- [Seats](/help/hub/seats/) — how seats work and how to add more.\n- [Billing and invoices](/help/hub/billing/) — download the invoices your details appear on.\n- [Credits](/help/hub/credits/) — what the one-time credit is worth and how it's spent.\n","path":"/help/hub/business-details/","section":"Your account","slug":"hub/business-details","title":"Add your business details","url":"https://bumbleb.co/help/hub/business-details/"},{"description":"Upload the data you already have — orders, listings, a log — and ask about it in plain words.","gate":null,"markdown":"\nYour revenue looks flat this quarter, and you suspect the real story is underneath: more customers\nspending less, or fewer spending more. You don't need website tracking to find out. Crunch isn't\nonly for clickstream — the trail of clicks and page views a website records. An orders export, a\nlistings feed or an operations log works the same way: you upload it, say what you need to know,\nand Crunch works out which columns can answer.\n\nThis guide walks you from a file on your computer to a first useful answer.\n\n## What you'll need\n\n- **A file with rows and columns** — Excel, CSV or Parquet. If you can open it as a spreadsheet,\n  it will almost certainly work.\n- **A BumbleB account.** The Free plan is enough to start —\n  [sign up](https://hub.bumbleb.co/signup) if you haven't.\n\nThat's all. There's no tag to install and no connector to set up. Each file you upload becomes a\n**dataset**: a table that lives in your own workspace.\n\n## Step by step\n\n1. **Export the data** from wherever it lives today — your store's admin, your booking system, a\n   shared spreadsheet. Most tools have an export or download option that gives you a CSV.\n2. **Upload it** from the **datasets** pill in the message box, with **Upload a file…**.\n   [Uploading files](/help/connect/upload-files/) covers size limits and what happens when you\n   upload the same file again.\n3. **Make sure the dataset is ticked** in the pill, then ask your question in plain words, the way\n   you'd ask a colleague who has the file open.\n4. **Tell Crunch what the file can't tell it** (see the next section). Do this once, at the start —\n   Crunch keeps it for the rest of the conversation.\n5. **Ask the next small question.** One step at a time gets you further than one big question —\n   see [Solve a problem in a conversation](/help/getting-started/solve-a-problem/).\n\n## Tell Crunch what it can't see\n\nA file carries values, not meaning. A rating column full of 0s doesn't say whether 0 means\n\"terrible\" or \"not rated\". So say once, at the start, anything a stranger opening your file would\nget wrong:\n\n- **What one row is** — \"each row is one order line, not one order\". This changes every count.\n- **Special values** — \"a rating of 0 means not rated\", \"-1 means unknown\". Left unsaid, they look\n  like real numbers and get averaged in.\n- **Codes and currencies** — \"country code 1 is India; amounts are in rupees\".\n\n## How to tell it worked\n\nWhen the upload finishes loading, the dataset is selected for the conversation and you can ask\nstraight away. Your first answer leads with the result and then shows its working, including\n**How I read your question** and **How this was counted**. If both match what you meant, you're\nset up. If not, say so in one line and ask again —\n[Read an answer](/help/getting-started/reading-an-answer/) explains each part.\n\n## An example\n\nA store's revenue looks flat, and the owner doesn't believe it's that simple. They have one orders\nexport and nothing else.\n\n| You ask | Crunch finds |\n|---|---|\n| Revenue was flat quarter over quarter. What changed underneath — the number of orders, or how big they were? | Revenue was $1.20M in both quarters, but orders rose from 8,000 to 8,800 while the average order fell from $150 to $136.36 — more orders, smaller ones. |\n\nA flat total was hiding two moves in opposite directions. That points straight at the next\nquestion: why did orders get smaller? From there, split it the way your business is organized — by\nregion, by channel, by product — one small step at a time.\n\n## Good to know\n\n- **Crunch reads what the file holds.** If the columns can't answer a question, it says so plainly\n  and tells you what would answer it, rather than guessing.\n- **Newer numbers mean a fresh upload.** An uploaded file is a snapshot. When you want this week's\n  orders, export and upload again, and ask in a new conversation — a conversation keeps reading\n  the version it started with.\n- **Two files answer together when they share a column** — an order ID, a customer ID, a date. See\n  [Ask one question across two datasets](/help/use-cases/two-datasets/).\n\n**Tip:** if a column name only makes sense to your team, explain it in your first message. It\ncosts one sentence and saves a wrong answer.\n\n## Related\n\n- [Uploading files](/help/connect/upload-files/) — formats, size limits and versions.\n- [Solve a problem in a conversation](/help/getting-started/solve-a-problem/) — how to break a big\n  question into small ones.\n- [Read an answer](/help/getting-started/reading-an-answer/) — what each part of an answer tells\n  you.\n- [Ask one question across two datasets](/help/use-cases/two-datasets/) — when one file isn't\n  enough.\n","path":"/help/use-cases/any-dataset/","section":"Use-case guides","slug":"use-cases/any-dataset","title":"Ask about any dataset","url":"https://bumbleb.co/help/use-cases/any-dataset/"},{"description":"Upload the market files you have the right to use, and ask what actually happened.","gate":null,"markdown":"\nYou've read that insiders were buying a particular stock this spring, and you'd like to see it for\nyourself in the filings rather than take a headline's word for it. Exchange files, regulatory\nfilings, positions from your broker or data vendor — upload them and ask across them in plain\nwords. Crunch describes what happened in the data. It does not give investment advice.\n\n## What you'll need\n\n- **The files**, from wherever you have the right to use them: a regulator's public data sets, an\n  exchange's daily files, your vendor's exports.\n- Each file in **CSV, Excel or Parquet**. Each one becomes its own dataset, a table in your\n  workspace.\n\n## Step by step\n\n1. **Check the source's terms.** Some exchange data can't be redistributed or loaded into\n   third-party tools. Public regulatory data usually can.\n2. **Download the files** you need.\n3. **Upload each one** — follow [Uploading files](/help/connect/upload-files/).\n4. **Pick them together** in the **datasets** pill, so one question can read across them. Crunch\n   works out how they line up — by date, by symbol, by filing.\n5. **Tell Crunch what it can't see** (below), then ask.\n\n## Tell Crunch what it can't see\n\nMarket files are full of codes and conventions that the people who publish them take for granted.\nSay once, at the start, what a newcomer would misread:\n\n- **Codes** — \"transaction code P is an open-market purchase\".\n- **Who counts** — \"insiders means Directors and Officers as filed\".\n- **Symbols** — how a symbol in one file maps to a company in another, if the names differ.\n- **Prices** — whether a price file is adjusted for splits and dividends, since an unadjusted price\n  jumps on the day of a split.\n- **The window** — give plain dates. Trading days are only the days in the files.\n\n## How to tell it worked\n\nEach upload finishes loading as its own dataset, selected for the conversation. In the answer,\n**How this was counted** lists the codes, roles and dates Crunch actually used — check them\nagainst the definitions you gave before you rely on a count.\n\n## An example\n\nThree public SEC files: 108,302 Form 4 filings, 114,297 insider rows (an anonymous ID and the\nfiled role — no names) and 179,708 transactions. A Form 4 is the filing a company insider makes\nwhen they buy or sell the company's shares.\n\n| You ask | Crunch finds |\n|---|---|\n| Counting only purchases (SEC transaction code P) dated January 1 to May 31, 2026, by insiders who filed as Director, Officer, or Director,Officer — which five company-months had the most different insiders buying, and how many in each? | Taiwan Semiconductor in April and in May (28 each), F\u0026M Bank Corp in March (15), Everforth in April (13) and Vital Farms in May (11). |\n| For Taiwan Semiconductor over the whole January–May period, how many different insiders bought, and on how many different days? | 29 different insiders, on 9 different days. |\n\nNotice how the first question spells out the code, the roles and the dates. That's what makes the\ncount exact. The follow-up then shows whether the buying was spread out or bunched together: 29\npeople over 9 days.\n\nSource: U.S. Securities and Exchange Commission, Insider Transactions Data Sets (public domain).\nA description of what happened — not investment advice.\n\n## Good to know\n\n- **Crunch works on the files you upload.** It does not fetch market data for you, so newer data\n  means a fresh download and upload.\n- **No investment advice.** It says what the data shows, not what to buy or sell.\n- **Counting \"different\" people or days is its own question.** \"How many purchases\" and \"how many\n  different insiders bought\" give different numbers. Say which one you mean.\n\n## Related\n\n- [Uploading files](/help/connect/upload-files/) — formats, size limits and versions.\n- [Ask one question across two datasets](/help/use-cases/two-datasets/) — how files line up.\n- [Solve a problem in a conversation](/help/getting-started/solve-a-problem/) — one small question\n  at a time.\n","path":"/help/use-cases/market-data/","section":"Use-case guides","slug":"use-cases/market-data","title":"Ask about market data","url":"https://bumbleb.co/help/use-cases/market-data/"},{"description":"The Analyst agent answers questions about your data, and shows you how it counted.","gate":null,"markdown":"\nThe Analyst is the agent you talk to by default. You type a question about your data in your own\nwords; it works out what to count, reads the data, and answers. Say you want to know whether last\nmonth's visits really went up — you ask, and you get a figure plus a plain account of how that\nfigure was counted, so you can decide whether to rely on it.\n\n## What you'll need\n\n- Some data to ask about — see [Datasets](/help/features/datasets/) for how to pick what a\n  conversation reads.\n\n## Asking a question\n\n1. If you want Crunch to work harder on this one, set the effort first with the brain button (see\n   **Thinking and effort** below).\n2. Type your question in the message box, in your business words. One question per turn works best.\n3. Send it, and read the answer from the top down — its parts are explained next.\n\n## Reading an answer\n\nAn answer can carry these parts, always in this order:\n\n- **The lead** — the answer in a sentence or two. After each figure there's a **↗**; click it to open\n  that figure's table or chart in the side panel.\n- **⚑ Check before relying on this** — anything the data could not confirm. It sits right under the\n  lead, so you see it before you act on the figure.\n- **How I read your question** — when a word could mean two things, this says which reading Crunch\n  took, for example \"I read *visits* as sessions, not page views\". Crunch never stops to ask you; it\n  picks the likelier reading and tells you, so you can reply \"answer it the other way\" if that isn't\n  what you meant.\n- **How this was counted** — the definitions it held and the rules it used for edge cases, written\n  from what it actually ran.\n- **Not established** — any part of your question it could not answer, with why and what would\n  answer it. Reply \"yes\" to have it try.\n- **Where this sits** — what is still open, or the natural next question.\n\n## Thinking and effort\n\nThe brain button beside the microphone sets how hard Crunch works on your next question:\n\n- **Thinking** (on by default) — Crunch reasons step by step before answering.\n- **Low effort** (the default) answers directly. **Medium** breaks the question down and checks its\n  work. **High** does the same with more reasoning, and uses more\n  [credits](/help/hub/credits/) — the units your usage is counted in.\n- **Time limit** — by default 5, 8 or 20 minutes for low, medium or high effort. When the time is up,\n  Crunch asks \"Do you want me to keep going on what's still open?\" Reply \"keep going\" to let it\n  continue. **No time limit** never pauses, which also means it can use many credits.\n- **Show every step** — shows each step of the work, instead of one live status line.\n\nThese settings are kept per conversation.\n\n## How to tell it worked\n\nYou get a lead with a figure and its **↗**, and a **How this was counted** section that matches what\nyou meant. If the reading or the counting isn't what you had in mind, say so in your next message.\n\n## Good to know\n\n- **Tip:** Say what a value means once, up front (\"a *trial* is anyone on the free plan\"). Crunch\n  keeps it for the rest of the conversation, so you don't have to repeat yourself.\n- **Tip:** Keep to one question per turn. A turn that asks one thing is far easier to check than one\n  that asks three.\n- You can also ask a product question, such as \"how do I upload a file?\". Crunch answers it from\n  these help pages.\n\n## Related\n\n- [Solve a problem in a conversation](/help/getting-started/solve-a-problem/) — how to phrase and\n  follow up a question.\n- [Charts, tables and artifacts](/help/features/artifacts/) — what you can do with each **↗**.\n- [Plan agent](/help/features/plan-agent/) — for a question that is really several questions.\n- [Choosing a model](/help/features/models/) — how the model and effort affect your usage.\n","path":"/help/features/analyst/","section":"Features","slug":"features/analyst","title":"Ask Crunch about your data","url":"https://bumbleb.co/help/features/analyst/"},{"description":"Ask across your website, your spend, your orders and your database at once.","gate":null,"markdown":"\nYour ad platform knows what you spent. Your website knows who signed up. Neither one alone can tell\nyou what a sign-up cost — and that's usually the number you actually need. Everything you bring\ninto BumbleB lands in one workspace, so a single question can span two datasets: website behavior\nwith ad spend, store events with orders, your database with your traffic. Crunch works out how the\ntwo relate before it answers.\n\n## What you'll need\n\n- **Two datasets in your workspace.** A dataset is one table of your data: events from your site\n  or store, a file you uploaded, or a table you linked. Not sure which route fits your data? See\n  [Which route for which data](/help/connect/which-route/).\n- **A column they share** — a campaign name, an order ID, a hashed email. This is how Crunch knows\n  which row in one dataset belongs with which row in the other.\n\n## Step by step\n\n1. **Find the column they share.** Look at both datasets and pick the one thing they have in\n   common. For ad spend and website visits it's usually the campaign name; for orders and\n   inventory it's the product.\n2. **Name it the same way in both**, if you can. If your export says `campaign` and the site says\n   `utm_campaign`, rename the column in the export before you upload it.\n3. **Say how they relate**, once, at the start of the conversation — for example: \"listings-search\n   in the spend export is the same campaign as utm_campaign on the site\". This covers values that\n   are spelled differently in the two places, which renaming a column can't fix.\n4. **Pick both datasets** in the **datasets** pill before you ask. A conversation only reads the\n   datasets it was given, so one that isn't ticked is invisible to it.\n5. **Ask your question** as if it were one table — \"what did each sign-up cost, by campaign?\"\n\n## How to tell it worked\n\nThe answer puts figures from both datasets side by side — spend from one, sign-ups from the other.\n**How this was counted** under the answer shows the rules it used, so you can check the two were\nmatched the way your team would match them. If a campaign is missing, it's usually spelled\ndifferently in the two places: tell Crunch they're the same and ask again.\n\n## An example\n\nTwo ad campaigns got the same budget, and the team suspects the cheap clicks aren't the good ones.\n\n| You ask | Crunch finds |\n|---|---|\n| We spent the same on our Google and Meta campaigns. How many sign-ups did each bring in, and what did each sign-up cost? | $12,000 on each. Google brought 960 sign-ups at $12.50 each; Meta brought 480 at $25.00 each. |\n\nThat answer joins the ad platforms' spend export to the site's sign-up events on the campaign name.\nNeither dataset could give it alone: one has the spend, the other has the sign-ups. The full setup\nis in [See which ad spend comes back](/help/use-cases/website-ad-roi/).\n\n## Good to know\n\n- **Once a conversation has read its data, its datasets are fixed** for that conversation, so\n  answers don't shift under you. To add another dataset, start a new conversation.\n- **Hashed emails line up people across sources** without the email itself ever being sent. See\n  [Hashing user IDs](/help/connect/hash-user-ids/).\n\n**Tip:** when two datasets cover different date ranges, say which window you want. Otherwise\n\"last month\" might mean a month one of them doesn't have.\n\n## Related\n\n- [See which ad spend comes back](/help/use-cases/website-ad-roi/) — this example, set up from\n  scratch.\n- [Datasets](/help/features/datasets/) — picking the datasets a conversation reads.\n- [Hashing user IDs](/help/connect/hash-user-ids/) — line up your site and your database on a\n  hashed email.\n","path":"/help/use-cases/two-datasets/","section":"Use-case guides","slug":"use-cases/two-datasets","title":"Ask one question across two datasets","url":"https://bumbleb.co/help/use-cases/two-datasets/"},{"description":"See every payment, download your invoices, and change the card you pay with.","gate":null,"markdown":"\nSooner or later someone asks for an invoice — your accountant at month end, or finance when they\nreconcile the card statement. The **Billing** page in the hub keeps all of that in one place, so you\ncan find a past payment, download its invoice, and see what's coming next without digging through\nemail.\n\n## What you'll find on the Billing page\n\n**Billing** lists every payment you've made. Each row shows:\n\n- the **date** of the payment;\n- a **description** of what it was for;\n- the **amount**;\n- its **status**;\n- a **Download** link for the invoice — the formal record of the payment that you can file or pass\n  to whoever handles your accounts.\n\nThe next upcoming charge is shown too, so you know what to expect before it happens.\n\n## Change the card you pay with\n\nIf your card is expiring, or you'd rather pay with a different one:\n\n1. Open **Subscriptions** in the hub.\n2. Go to **Payment Method** and click **Update**.\n3. Enter the new card details.\n\nPayments go through Razorpay, the payment provider Crunch uses, so the card form you see comes from\nthem.\n\n## Where the other billing details live\n\nTwo other pieces of information end up on, or around, your invoices, and each has its own home:\n\n- **Your billing contact and address** — the person who receives billing information and the\n  address the bill is made out to — are under **Subscriptions**.\n- **Your company details** — business name, tax ID and so on — are under\n  [Business details](/help/hub/business-details/).\n\n**Tip:** If an invoice needs your company name or tax number on it, fill in\n[Business details](/help/hub/business-details/) first. That page explains when the details start\nappearing on invoices.\n\n## Related\n\n- [Business details](/help/hub/business-details/) — put your company's details on your invoices.\n- [Upgrading, downgrading and cancelling](/help/hub/subscription/) — change what you pay for, and\n  when a change is charged.\n- [Credits](/help/hub/credits/) — buy more usage when your included usage runs out.\n","path":"/help/hub/billing/","section":"Your account","slug":"hub/billing","title":"Billing and invoices","url":"https://bumbleb.co/help/hub/billing/"},{"description":"No ad-platform connector yet? Export your spend from Google Ads, Meta or any platform and upload it.","gate":null,"markdown":"\nYou want to know what a real sign-up cost you, campaign by campaign, not just what a click cost.\nTo answer that, Crunch needs your ad spend sitting next to what visitors did on your site. There\nare no direct ad-platform connectors yet, but you don't need one: an export from each platform is\nall it takes.\n\nAn export is the file an ad platform lets you download from its reports, usually a CSV or Excel\nsheet with one row per day and campaign.\n\n## What you'll need\n\n- Access to the reports in each ad platform you spend on (Google Ads, Meta, or any other).\n- A way to upload a file to Crunch — see [Uploading files](/help/connect/upload-files/).\n- Your site's events, if you want to ask about sign-ups or purchases — see\n  [Google Tag Manager tag](/help/connect/google-tag-manager/).\n\n## Step by step\n\n1. **Export your spend by day and by campaign.** In each ad platform, download a report with the\n   date, campaign, spend, clicks and impressions. Daily rows let you ask about any window later.\n2. **Name the campaign column the way your site's events name it.** If your ad links carry\n   `utm_campaign`, call the export's column `utm_campaign` too. Two datasets only line up on a\n   column they share, and differently named keys don't line up.\n3. **Upload the file.** Follow [Uploading files](/help/connect/upload-files/). If two platforms\n   export different layouts, upload them as two separate files — that's fine.\n4. **Ask across both.** Pick your spend and your site's events, then ask something like \"what did\n   a sign-up cost per campaign?\" The [Website \u0026 ad ROI](/help/use-cases/website-ad-roi/) guide walks\n   through a full example.\n\n## Good to know\n\n- **Heads up:** an export is a snapshot. When you want newer spend, upload a fresh export as a new\n  version of the same dataset, rather than a brand-new one.\n- **Tip:** if the campaign is called `campaign` in the export and `utm_campaign` on your site,\n  rename one before you upload. A campaign named differently in the two places won't line up\n  until you do.\n\n## Related\n\n- [Website \u0026 ad ROI](/help/use-cases/website-ad-roi/) — the full guide, with a worked example.\n- [Two datasets, one question](/help/use-cases/two-datasets/) — how datasets line up.\n- [Which route for which data](/help/connect/which-route/) — every way to bring data in.\n","path":"/help/connect/ad-spend/","section":"Connect your data","slug":"connect/ad-spend","title":"Bring in your ad spend","url":"https://bumbleb.co/help/connect/ad-spend/"},{"description":"Open, copy, download and trace every table and chart an answer draws.","gate":null,"markdown":"\nEvery figure in an answer has a table or chart behind it, and you can open it, take it with you, or\ntrace it back to where it came from. That matters when you want to paste a chart into a deck, check\nthe rows yourself before you quote a number, or hand a colleague the exact query that made it.\n\n## What you'll need\n\n- An answer with at least one figure — see [Ask Crunch about your data](/help/features/analyst/).\n- For the data behind a figure (**⛁**): [Data home](/help/features/data-home/), which is on Team\n  plans (see [pricing](/pricing/)).\n\n## Open, copy or download a figure\n\n1. In the answer, click the **↗** after a figure. Its table or chart opens in the side panel.\n2. To take it with you, use **Copy** or **Download** in the panel header.\n\n## See the data behind a figure\n\n1. Click the **⛁** beside a **↗**.\n2. You'll see where the data came from, its rows a page at a time, the SQL that made it, and a box\n   to run your own SQL on it.\n\nTo look at the whole dataset a chart was drawn from, click **Explore data** in the chart's panel\nheader.\n\n## Good to know\n\n- **Tip:** If you want every chart from a conversation in one place, the\n  [Report agent](/help/features/report-agent/) gathers a conversation's charts into one document.\n\n## Related\n\n- [Data home and lineage](/help/features/data-home/) — browse datasets and walk a chart back to its\n  source.\n- [Report agent](/help/features/report-agent/) — turn a conversation into a document.\n- [Ask Crunch about your data](/help/features/analyst/) — how to read an answer and its **↗** links.\n","path":"/help/features/artifacts/","section":"Features","slug":"features/artifacts","title":"Charts, tables and artifacts","url":"https://bumbleb.co/help/features/artifacts/"},{"description":"Pick the model for your next message, and see what it does to your usage.","gate":null,"markdown":"\nCrunch can answer with different AI models. A more capable model is worth it for a hard question; a\nlighter one makes your usage go further on everyday ones. You choose per message, so you can save\nthe heavy model for the questions that need it.\n\n## Choose a model\n\n1. Find the model chip (the chip icon) in the message box. It shows the model for your **next**\n   message.\n2. Click it and choose from the models your plan includes.\n\nYou can switch mid-conversation; the conversation's history is kept.\n\n## What it does to your usage\n\nEach model shows how much usage a message draws compared with your plan's standard model — for\nexample **0.2×** or **15×**. More capable models use more; lighter ones stretch your usage further.\n\nEffort matters too: **High effort** and **No time limit** use more. See\n[Thinking and effort](/help/features/analyst/) for what those settings do.\n\n## Related\n\n- [Usage](/help/hub/usage/) — see what you've used so far.\n- [Credits](/help/hub/credits/) — how usage is counted.\n- [Long conversations](/help/features/long-conversations/) — the context meter.\n","path":"/help/features/models/","section":"Features","slug":"features/models","title":"Choosing a model","url":"https://bumbleb.co/help/features/models/"},{"description":"A custom pixel in your store's Customer events — no theme edit, no app, no developer.","gate":null,"markdown":"\nTo ask why sales moved, Crunch needs to see what shoppers do in your store, not just what they buy.\nThe BumbleB pixel sends your storefront's standard events — page and product views, add to cart,\ncheckout started and checkout completed — to your workspace. A pixel is a small piece of code\nShopify runs for you on every page. You add it in your Shopify admin, with no theme edit, no app\nand no developer.\n\nThe pixel asks Shopify for no customer identity, so it sends what shoppers did, not who they are.\n\n## What you'll need\n\n- Access to the **Connectors** page in the hub — see [Roles](/help/hub/roles/).\n- Access to your Shopify admin.\n- Your store's own domain (like `yourstore.com`) and access to its DNS settings, to prove the store\n  is yours. A `.myshopify.com` address can't carry the record this needs, so if you have no custom\n  domain, [contact us](/help/trouble/contact-support/).\n\n## Step 1: Create the connector\n\n1. In the hub, open **Connectors** and click **+ Create connector**.\n2. Set **Source** to Shopify.\n3. Give it a **Connector Name**. The **Stream ID** fills itself in.\n4. Enter your store's own domain as the **Website Domain**.\n5. Click **Create** and copy the API key.\n\n**Heads up:** the key is shown **once**. Paste it somewhere safe straight away. If you lose it,\nopen the connector card and click **+ Generate New API Key** under **API Keys** for a new one.\n\n## Step 2: Verify your domain\n\nFollow [Verifying your domain](/help/hub/domains/). Events are only accepted once it verifies, so\nit's worth doing before you add the pixel.\n\n## Step 3: Add the pixel in Shopify\n\n1. In your Shopify admin, go to **Settings → Customer events**.\n2. Click **Add custom pixel** and name it `BumbleB`.\n3. Paste the code from the connector card, replacing `YOUR_API_KEY_HERE` with your key.\n4. Click **Save**, then **Connect**.\n\n## How to tell it worked\n\nCheck progress on the connector card's **Pipeline Status**, which names any error by stage, with\nwhat to do.\n\n## Good to know\n\n- **Events start on the day you connect the pixel.** There is no history before that. For earlier\n  orders, upload your orders export — see [Uploading files](/help/connect/upload-files/).\n- **Trends need time.** Questions about trends need a few weeks of collection first, so the sooner\n  the pixel is in, the sooner you can ask them.\n\n## Related\n\n- [Shopify \u0026 e-commerce guide](/help/use-cases/shopify/) — what to ask once the pixel is in.\n- [Uploading files](/help/connect/upload-files/) — bring in orders, inventory and refunds.\n- [Fixes](/help/connect/fixes/) — when events aren't arriving.\n","path":"/help/connect/shopify-pixel/","section":"Connect your data","slug":"connect/shopify-pixel","title":"Connect your Shopify store with the pixel","url":"https://bumbleb.co/help/connect/shopify-pixel/"},{"description":"Send your website's or web app's events to Crunch through one Custom HTML tag.","gate":null,"markdown":"\nCrunch answers questions about what people do on your site — where they drop off, which campaign\nbrings sign-ups — from your site's events. An event is one thing a visitor did, like viewing a page\nor signing up. If your site already uses Google Tag Manager, you can send those events with one\nCustom HTML tag, and you stay in control: only the events you push are sent.\n\nHere is how it fits together. Your site pushes events to the data layer — a small list in the page\nthat Google Tag Manager watches — and the BumbleB tag in your container passes them on to your\nworkspace.\n\n## What you'll need\n\n- A Google Tag Manager container on your site, and access to edit it.\n- The **Connectors** page in the hub, with a role that can manage connectors — see\n  [Roles](/help/hub/roles/).\n- Access to your domain's DNS settings (usually at the company you bought the domain from), to\n  prove the site is yours.\n\n**Heads up:** events are only accepted once a domain is verified, and events sent before that are\nnot kept. Start the domain step early, since the internet can take a while to notice the change.\n\n## Step 1: Create the connector\n\nA connector is the link between one source of data and your workspace.\n\n1. In the hub, open **Connectors** and click **+ Create connector**.\n2. Set **Source** to Google Tag Manager.\n3. Give it a **Connector Name**, for example \"Production Website\".\n4. Check the **Stream ID**. It is filled in from the name, and must be 3–50 characters, start with a\n   letter, and use only lowercase letters, digits and `_`.\n5. Enter your **Website Domain**, for example `example.com`.\n6. Click **Create**. Your API key is shown **once**, so click **Copy Key** and keep it somewhere\n   safe — you'll paste it into the tag in Step 3.\n\n## Step 2: Verify your domain\n\nFollow [Verifying your domain](/help/hub/domains/): you publish one TXT record at your registrar and\nclick **Check again**. Once it verifies, the connector starts receiving data.\n\n## Step 3: Set up Google Tag Manager\n\nOn the connector's card, the **GTM Setup Guide** shows three parts. Create each one in your GTM\ncontainer:\n\n1. A **Trigger** of type *Custom Event*, with the event name `bb_event`. This tells the tag when to\n   fire.\n2. A **Variable** of type *Data Layer Variable*, named `bb`. This is where the tag reads your event\n   from.\n3. A **Tag** of type *Custom HTML*, fired by that trigger. Copy the code from the card, replace\n   `YOUR_API_KEY_HERE` with your key, and publish the container. Nothing goes live until you\n   publish.\n\n**Heads up:** use **one connector per GTM container**. Two BumbleB tags in one container send every\nevent twice.\n\n## Step 4: Push your events\n\nFrom your site's code, push each event you want tracked. This snippet sends a sign-up, along with\nthe page it happened on:\n\n```js\ndataLayer.push({\n  event: 'bb_event',\n  bb: { event_name: 'signup', page_url: location.href }\n});\n```\n\n`event_name` is the only required field. Good first events are `page_view`, `signup`, `purchase`,\n`feature_used` and `error` — the moments you'll actually ask about.\n\n**Tip:** keep the `utm_` parameters on your ad links. They carry the campaign name, which is how\nCrunch later ties a visit back to the ad that brought it.\n\n## How to tell it worked\n\n- The connector card shows **Active** once the domain is verified.\n- Each API key shows **Last used**. It only updates when an event is accepted, so a fresh date means\n  events are getting through.\n- **Pipeline Status** on the card lists any errors by stage, with what to do.\n- If an event is refused, your browser's console logs `[BumbleB] event refused`.\n\n## Related\n\n- [Hashing user IDs](/help/connect/hash-user-ids/) — if your events carry an email or phone number.\n- [Which route for which data](/help/connect/which-route/) — other ways to bring data in.\n- [Fixes](/help/connect/fixes/) — when events aren't arriving.\n- [Product \u0026 behavior](/help/use-cases/product-behavior/) — what to ask once events flow.\n","path":"/help/connect/google-tag-manager/","section":"Connect your data","slug":"connect/google-tag-manager","title":"Connect your website with Google Tag Manager","url":"https://bumbleb.co/help/connect/google-tag-manager/"},{"description":"Create the connectors that send your data to Crunch, and check they're working.","gate":null,"markdown":"\nCrunch can only answer questions about data it receives. A **connector** is the link that sends\nthat data in — from your website, your Shopify store or a database in your own network. The\n**Connectors** page in the hub is where you create one and, later, where you check it's healthy if\nthe numbers ever look off.\n\n## What you'll need\n\n- Access to the hub — see [Roles](/help/hub/roles/) for who can set up connectors.\n- For a website or Shopify store, a domain you can verify. Connectors for those two sources need a\n  [verified domain](/help/hub/domains/) before they work.\n- For an on-prem database, a Team plan — see [pricing](/pricing/).\n\n## Step by step\n\n1. In the hub, open **Connectors**.\n2. Click **+ Create connector**.\n3. Choose the source that matches where your data lives:\n   - **Google Tag Manager** — for your website or web app.\n     [Set it up](/help/connect/google-tag-manager/).\n   - **Shopify** — for your store. [Set it up](/help/connect/shopify-pixel/).\n   - **On-prem database** (Team) — a container (a small packaged program) that runs in your own\n     network. [Set it up](/help/features/on-prem-sync/).\n4. Follow the setup guide for that source, linked above and on the connector's card.\n\n## How to tell it worked\n\nEach connector has a card. Its status reads either:\n\n- **Awaiting domain verification** — the connector exists, but its domain hasn't been verified yet;\n  see [Verifying your domain](/help/hub/domains/). Events sent before then are refused and not\n  kept, so you can install the tag or pixel early, but data starts only once a domain verifies.\n- **Active** — the connector is ready. For a website or Shopify connector, this is the point where it\n  starts receiving data.\n\nThe card also shows:\n\n- the connector's **setup guide**;\n- its **API keys** — the secret codes that identify your connector — with **Last used**, so you can\n  see when each key last did anything (website and Shopify connectors);\n- its **Verified Domains**, with a link to manage them (website and Shopify connectors);\n- **Pipeline Status** — any errors, grouped by stage, with what to do about each.\n\n**Tip:** If your numbers look thin or stop updating, look at **Pipeline Status** first. It tells you\nwhere things went wrong and what to do about it.\n\n## Good to know\n\n- **Use one connector per GTM container.** A GTM container is the bundle of tags Google Tag Manager\n  runs on your site.\n- **Website and Shopify connectors need a verified domain.** Do that early — see\n  [Verifying your domain](/help/hub/domains/).\n\n## Related\n\n- [Verifying your domain](/help/hub/domains/) — the step that lets a website or Shopify connector\n  become Active.\n- [Connect with Google Tag Manager](/help/connect/google-tag-manager/) — the full website setup.\n- [Connect your Shopify store](/help/connect/shopify-pixel/) — the full store setup.\n- [On-prem sync](/help/features/on-prem-sync/) — sync a database in your own network.\n","path":"/help/hub/connectors/","section":"Your account","slug":"hub/connectors","title":"Connectors in the hub","url":"https://bumbleb.co/help/hub/connectors/"},{"description":"How to reach us — from the chat or by email.","gate":null,"markdown":"\nIf something in Crunch isn't working, or you have a question these pages don't answer, here's how to\nreach us. The quickest route depends on what you need.\n\n## From the chat\n\nClick **Report a bug** in the header. We reply through the bell — your\n[notifications](/help/features/notifications/). This is the best route when something went wrong in a conversation — see\n[Report a bug](/help/features/report-a-bug/).\n\n## By email\n\n- **support@bumbleb.co** — for customers who need help.\n- **reach@bumbleb.co** — for anything else.\n\n## Related\n\n- [Slow, empty or \"not established\" answers](/help/trouble/answer-problems/) — quick fixes to try\n  first.\n- [Report a bug](/help/features/report-a-bug/) — what to include when you report a problem.\n","path":"/help/trouble/contact-support/","section":"Trouble \u0026 contact","slug":"trouble/contact-support","title":"Contact support","url":"https://bumbleb.co/help/trouble/contact-support/"},{"description":"When Crunch reads your question differently from what you meant, one line puts it back on track.","gate":null,"markdown":"\nBusiness words are often ambiguous. \"Visits\" could mean sessions or page views; \"orders\" could mean\nwhole orders or the individual lines inside them. Crunch has to pick a reading, and sometimes it picks\na different one from the one you had in mind. You don't need to start over when that happens — you\ncorrect it in one line, the way you'd correct a colleague.\n\n## How you'll know it misread you\n\nEvery answer says how Crunch read your question. Read that part first: if the reading isn't what you\nmeant, the figures that follow answer a different question from yours, however tidy they look. See\n[Reading an answer](/help/getting-started/reading-an-answer/) for where that part sits in an answer.\n\n## Step by step\n\n1. **Find how Crunch read your question** in the answer, and check it against what you meant.\n2. **Reply with the correction in one line.** Say what to use instead, in your own business words.\n3. **Ask again**, and read the new answer's reading to confirm it took the correction.\n\n## What a correction sounds like\n\nShort and specific works best. A few examples:\n\n- \"Use the app the enquiry came from, and include web.\"\n- \"By visits I mean sessions, not page views.\"\n- \"Count orders, not order lines.\"\n- \"Leave out our office traffic.\"\n\nEach one names the thing to change and what to change it to, so there's nothing left to guess.\n\n## Good to know\n\n- **A correction lasts for the conversation.** Crunch keeps it for the rest of the conversation, so\n  you don't have to repeat it with every question.\n- **A new conversation starts fresh.** If you find yourself making the same correction again and\n  again, say it once at the start of each new conversation — \"by visits I mean sessions\" — before you\n  ask your first question.\n\n## Related\n\n- [Reading an answer](/help/getting-started/reading-an-answer/) — what each part of an answer tells\n  you, including how Crunch read you.\n- [Solve a problem in a conversation](/help/getting-started/solve-a-problem/) — how to ask in small\n  steps so each answer sets up the next.\n- [Slow, empty or \"not established\" answers](/help/trouble/answer-problems/) — what to do when an\n  answer isn't what you need.\n","path":"/help/getting-started/correcting-crunch/","section":"Getting started","slug":"getting-started/correcting-crunch","title":"Correct Crunch when it misreads you","url":"https://bumbleb.co/help/getting-started/correcting-crunch/"},{"description":"What a credit is, when credits are spent, how to buy more, and what happens when they run out.","gate":null,"markdown":"\nEvery question you ask Crunch uses a little computing, and that use is measured. Your plan includes\nsome usage every month; **credits** are what pay for usage beyond that. Knowing how the two fit\ntogether means you won't be surprised mid-analysis — say, on the last week of the month when the\nteam is busiest.\n\n## What a credit is worth\n\n**1,000 credits = US $1** of usage, whatever currency you pay in. So the value of a credit is the\nsame whether you're billed in rupees or US dollars.\n\n## Which gets spent first\n\nEach month your plan includes usage for every seat — a seat is one person on your plan.\n\n1. That **included usage is spent first**.\n2. **Credits are spent after it**, once the included usage is used up.\n\nThe included usage resets on your billing date each month and does not roll over, so unused\nincluded usage from one month isn't added to the next.\n\n## Buy credits\n\n1. In the hub, open **Credits**.\n2. Click **Buy Credits** and choose a package.\n3. Pay through Razorpay, the payment provider Crunch uses.\n\nThe credits arrive once the payment completes.\n\n## Earn credits\n\nYou can also earn credits, once each, by:\n\n- [verifying a domain](/help/hub/domains/) — proving you own the site a connector collects from;\n- having your [business details](/help/hub/business-details/) approved.\n\n## How to tell what happened\n\n**Recent Activity** on the **Credits** page lists every grant, purchase and use. If your balance\nlooks different from what you expected, that list shows you each step that got it there.\n\n## When it runs out\n\nYou'll get warnings before you hit the limit, so it shouldn't catch you off guard:\n\n- At **75%** and **90%** of your included usage, the usage meter turns amber, then red.\n- At **90%**, and again when your included usage is used up, you also get a notification in the\n  bell and by email, with a link to buy credits. If you're on a teammate's Team plan, it says\n  whose credits pay for usage past your allowance.\n- When **both** your included usage **and** your credits are used up, new questions are refused\n  until you add credits or upgrade.\n\n**Tip:** When you see the 90% warning ahead of a busy stretch, buy credits then rather than waiting\nfor questions to be refused.\n\n## Related\n\n- [Your usage](/help/hub/usage/) — see how much of this month's included usage you've used.\n- [Choosing a model](/help/features/models/) — bigger models and higher effort use more per\n  question.\n- [Upgrading, downgrading and cancelling](/help/hub/subscription/) — the other way to get more\n  usage.\n","path":"/help/hub/credits/","section":"Your account","slug":"hub/credits","title":"Credits","url":"https://bumbleb.co/help/hub/credits/"},{"description":"Browse every dataset, walk any chart back to its source, and keep the SQL you run.","gate":"xylem","markdown":"\nData home is where everything Crunch has read or made for you lives in one place. A dataset here\nmeans any table Crunch can read — your clickstream (the events your website or store sends), a file\nyou uploaded, a table you linked, or a result a chat turn made. Lineage is the trail from your\nsources to a given dataset. Come here when a chart in a conversation surprises you and you want to\nsee exactly where its numbers came from, or when you want to run your own query and keep it.\n\n## What you'll need\n\n- A Team plan (see [pricing](/pricing/)).\n\n## Open it\n\nClick **Data** (the database icon) in the left navigation. You can also get here from an answer:\nclick **⛁** beside any figure, or **Explore data** on a chart.\n\n## Find your way around\n\n- **Analyses** — the datasets each chat turn made, grouped by turn. **Open in chat** takes you back\n  to the conversation.\n- **Datasets**, **Sources** (your clickstream, uploads and linked tables) and **Queries**.\n- Filter any list by name. Lists show 50 at a time.\n\n## Look at a dataset\n\nOpen a dataset to see:\n\n- its rows, with **Export CSV** to take them away;\n- its columns;\n- **Made by** — the SQL that made it;\n- **Lineage** — a graph from your sources to this dataset. Click a node to walk to it.\n\n## Run SQL\n\n1. Type what you want in words, such as \"sessions by device last week\", and click **Draft** — or\n   write the SQL yourself. In your SQL, `{{this}}` is the dataset you are on.\n2. Click **Check** before you **Run**.\n3. Results come 100 rows a page.\n\n**Tip:** Add an `ORDER BY` to your query, so the pages stay stable as you move through them.\n\n## Queries you ran\n\nEvery query you run is kept, so you can come back to it. **Star** one to keep it at the top, give it\na title, see each time it ran, and **Rerun** it.\n\n**Heads up:** If a source a saved query read is gone, the query says so, rather than quietly reading\nsomething else.\n\n## Related\n\n- [Charts, tables and artifacts](/help/features/artifacts/) — the **↗** and **⛁** links in an\n  answer.\n- [SQL console](/help/features/sql-console/) — ask in words and run SQL from inside a conversation.\n- [Datasets](/help/features/datasets/) — pick what a conversation reads.\n","path":"/help/features/data-home/","section":"Features","slug":"features/data-home","title":"Data home and lineage","url":"https://bumbleb.co/help/features/data-home/"},{"description":"Choose what a conversation reads, and which version of it.","gate":null,"markdown":"\nA dataset is a table Crunch can read — a file you uploaded, your clickstream (the stream of events\nyour site sends), or a table you linked. Before you ask, you tell the conversation which datasets to\nread, and which version of each. Once it has read them, they stay fixed, so a figure you got this\nmorning doesn't quietly change because new data arrived at lunch.\n\n## What you'll need\n\n- At least one dataset, or a file to upload — see [Upload files](/help/connect/upload-files/).\n\n## Pick datasets\n\n1. Click the **datasets** pill (the database icon) in the message box.\n2. **Search datasets** and tick the ones you want — or choose **Upload a file…** to add a new one.\n3. For a selected dataset, you can choose:\n   - **Use as clickstream** — read it as events, which is what funnels, retention and paths need.\n   - **Snapshots** — which version to read (see below).\n   - **Clear selection** — start the pick again.\n\n## Choose a version\n\n- **A table you replace with each upload.** Crunch keeps each upload as a version. You can read\n  **As first read** (the default), **Newest**, or a specific one.\n- **A table that grows over time.** You can read **All snapshots**, or a chosen set of them.\n\n## How to tell it worked\n\nOnce the conversation has read its data, the pill shows **In play (fixed for this conversation)**.\nFrom then on, its datasets and versions are fixed, and answers don't shift under you.\n\n## Good to know\n\n- **Heads up:** Because the pick is fixed once read, ask about new data in a new conversation.\n- Every upload, and every dataset a chat makes, lives in your own workspace.\n\n## Related\n\n- [Upload files](/help/connect/upload-files/) — add a dataset from a file.\n- [Your own data lake](/help/features/external-connections/) — read tables where they already live.\n- [Data home and lineage](/help/features/data-home/) — browse every dataset and where it came from.\n","path":"/help/features/datasets/","section":"Features","slug":"features/datasets","title":"Datasets: picking, pinning and versions","url":"https://bumbleb.co/help/features/datasets/"},{"description":"Your store's pixel plus the exports you already have — and why the sales moved.","gate":null,"markdown":"\nA dashboard tells you sales dropped. Knowing **why** — demand, stock, price, a broken checkout —\ntakes your store's events next to its orders and inventory. Was it fewer shoppers, or shoppers who\ncouldn't buy the thing they came for? With your store's events and a few exports side by side, you\ncan ask Crunch exactly that, and get an answer with the figures behind it.\n\n## What you'll need\n\n- **The BumbleB custom pixel** in your Shopify store. It records what shoppers do — product\n  views, add-to-carts, checkouts.\n- **Exports from Shopify** as your questions need them: orders (order lines), inventory levels,\n  products, refunds — and ad spend from your ad platforms. An export is the file Shopify or an ad\n  platform lets you download from its admin.\n\n## Step by step\n\n1. **Add the pixel** in your Shopify admin under **Settings → Customer events** — see\n   [Shopify pixel](/help/connect/shopify-pixel/) for each step.\n2. **Upload your orders export.** The pixel records from the day you install it, with no history\n   before that, so your orders export is what covers the past.\n3. **Upload inventory, products and refunds** when a question needs them — see\n   [Uploading files](/help/connect/upload-files/). You don't need them all on day one.\n4. **Pick the datasets** your question needs in the **datasets** pill, tell Crunch what it can't\n   see (below), and ask.\n\n## Tell Crunch what it can't see\n\nEvery store's exports have their own habits. Say once, at the start:\n\n- **What \"out of stock\" means in your inventory export** — \"available at 0 means out of stock\".\n- **Refunds versus orders** — whether a refund is its own row or a negative order line, so a refund\n  is never mistaken for a sale.\n- **Dates** — the windows you want compared, in plain dates.\n\n## How to tell it worked\n\nThe pixel's connector card in the hub shows its progress under **Pipeline Status**, which names\nany error by stage along with what to do. In the chat, your first answer should name the figures\nfor both windows you asked about, and **How this was counted** should match how your store counts\nrevenue.\n\n## An example\n\nWine sales felt slow in mid-September.\n\n| You ask | Crunch finds |\n|---|---|\n| Wine revenue fell from September 10 to 24 compared with August 18 to September 1. By how much, and which wine carried it? | Revenue fell from $10,664 to $5,774 — down $4,890. One wine carried it: the Pet-Nat Rosé went from $3,784 to nothing. |\n\nWhat felt like a slow fortnight turned out to be one wine dropping from $3,784 to nothing. The next\nquestion is why: was it out of stock, or did shoppers stop looking? Ask that next, with the\ninventory export picked.\n\n## Good to know\n\n- **Shopify today.** Other e-commerce platforms come in as exports, which work for every question\n  that doesn't need live shopper behavior.\n- **The pixel starts on the day you install it.** For anything earlier, use your orders export.\n- **Ask \"why\" one step at a time.** First find where the drop sits, then check stock, then views.\n  Each answer narrows the next question — see\n  [Solve a problem in a conversation](/help/getting-started/solve-a-problem/).\n\n## Related\n\n- [Shopify pixel](/help/connect/shopify-pixel/) — install it step by step.\n- [Uploading files](/help/connect/upload-files/) — bring in orders, inventory and refunds.\n- [See which ad spend comes back](/help/use-cases/website-ad-roi/) — put ad spend next to your\n  store's events.\n","path":"/help/use-cases/shopify/","section":"Use-case guides","slug":"use-cases/shopify","title":"Find out why your store's sales moved","url":"https://bumbleb.co/help/use-cases/shopify/"},{"description":"The snags people hit most often when connecting data, and what to do about each one.","gate":null,"markdown":"\nMost connection problems come down to a handful of causes: a domain that isn't verified yet, a tag\nthat isn't firing, a file Crunch can't read, or a conversation looking at the wrong data. Find the\nsymptom you're seeing below, and the fix is usually a minute's work.\n\n## No events are arriving\n\n- **The connector says \"Awaiting domain verification\".** Crunch accepts nothing from your site\n  until a domain is verified, which is how it knows the site is really yours. Finish the steps in\n  [Verifying your domain](/help/hub/domains/). Events sent before that are not kept, so they won't\n  appear later.\n- **Your browser console shows `[BumbleB] event refused`.** Check that the API key in the tag is\n  exactly the one the hub gave you, with no stray spaces, and that the page is on your verified\n  domain.\n- **Last used never changes.** That date only updates when an event is accepted, so if it never\n  moves, the tag isn't firing. In Google Tag Manager, check that the trigger is a Custom Event named\n  `bb_event`, the variable is named `bb`, and the container is published.\n- **Still stuck?** **Pipeline Status** on the connector card names any error by stage, and tells\n  you what to do about it.\n\n## Every event shows up twice\n\nTwo BumbleB tags in one Google Tag Manager container send everything twice. Use one connector per\ncontainer, and remove the extra tag.\n\n## An upload is rejected\n\n- **The file is too big.** Files over **100 MB**, or with more than **300 columns**, are turned\n  away. Split the file, or drop the columns you don't need.\n- **The format isn't supported.** Crunch reads `.csv`, `.tsv`, `.xlsx`, `.xls` and `.parquet`, and\n  one sheet per upload. Save other formats as one of these first.\n- **A date came through as text.** A date that could be read as day/month or month/day, like\n  03/04, stays text so it is never read the wrong way round. Write dates as `YYYY-MM-DD` before you\n  upload.\n\n## The answer used the wrong data\n\nA conversation reads the datasets that were picked when it started. If an answer drew on the wrong\ndata, pick the right datasets in the datasets pill and ask again in a new conversation — see\n[Datasets](/help/features/datasets/).\n\n## Related\n\n- [Google Tag Manager tag](/help/connect/google-tag-manager/) — set the tag up from the start.\n- [Uploading files](/help/connect/upload-files/) — upload steps and limits.\n- [Contact support](/help/trouble/contact-support/) — when none of these fit.\n","path":"/help/connect/fixes/","section":"Connect your data","slug":"connect/fixes","title":"Fix common problems: no events, rejected uploads, the wrong dataset","url":"https://bumbleb.co/help/connect/fixes/"},{"description":"Turn an email or phone number into a one-way hash in the visitor's browser, so the raw value never leaves the page.","gate":null,"markdown":"\nIf your site knows who a visitor is by their email address or phone number, you probably don't\nwant that value travelling anywhere. The BumbleB tag can hash it **in the visitor's browser**, so\nthe raw value never leaves the page.\n\nA hash is a one-way scramble: the same email always turns into the same string of characters, but\nyou can't turn the string back into the email. That means Crunch can still tell that two visits\ncame from the same person, without ever seeing who they are.\n\n## What you'll need\n\n- A Google Tag Manager connector that is already set up — see\n  [Google Tag Manager tag](/help/connect/google-tag-manager/).\n- Access to edit your Google Tag Manager container, because you'll paste the tag code again.\n\n## Step by step\n\n1. In the hub, open **Connectors** and find your GTM connector's card.\n2. Under the tag code, set **Hash user id in the page** to the option that matches your user ID:\n   - **No, send it as the site sets it** — for an internal ID that isn't personal, like an account\n     number.\n   - **Yes, it is an email** — the email is trimmed and lowercased, then hashed. Lowercasing first\n     means `Ana@Example.com` and `ana@example.com` hash the same way.\n   - **Yes, it is a phone number** — the number is reduced to `+` and digits, then hashed, so\n     spaces and dashes don't change the result.\n3. Copy the tag code again and **re-paste it** into Google Tag Manager, then publish the container.\n   The setting lives in the code itself, so changing it in the hub alone does nothing.\n\n## Good to know\n\n- The hash is SHA-256, a widely used standard.\n- **Heads up:** if a browser can't compute the hash — for example on a page not served over HTTPS —\n  the user ID is left out rather than sent raw.\n\n## When you need it\n\nTurn it on when you also sync a database that holds the same people. The\n[on-prem sync connector](/help/features/on-prem-sync/) can hash its email or phone column the same\nway, so the two line up without either side ever sending the raw value.\n\n## Related\n\n- [Google Tag Manager tag](/help/connect/google-tag-manager/) — the tag this setting belongs to.\n- [On-prem sync connector](/help/features/on-prem-sync/) — bring in a database with matching hashes.\n- [Two datasets, one question](/help/use-cases/two-datasets/) — ask across your site and your\n  database.\n","path":"/help/connect/hash-user-ids/","section":"Connect your data","slug":"connect/hash-user-ids","title":"Hash user IDs before anything is sent","url":"https://bumbleb.co/help/connect/hash-user-ids/"},{"description":"Add people to your Team plan, choose their role, and remove someone when they leave.","gate":null,"markdown":"\nCrunch is more useful when the people who ask the questions can ask them directly — your marketing\nlead, your ops manager, a co-founder — instead of routing everything through you. Inviting them\ngives each person their own access, and you choose what each of them can do.\n\n## What you'll need\n\n- **A Team plan.** Inviting is for Team plans; a Starter plan has one seat (one person). See\n  [pricing](/pricing/).\n- **A free seat for each person.** Members plus pending invites can't exceed the seats you have.\n  You can add seats under [Subscriptions](/help/hub/subscription/).\n- **Permission to invite.** See [Roles](/help/hub/roles/) for who can invite and manage users.\n\n## Step by step\n\n1. In the hub, open **Users** and click **+ Invite Member**.\n2. Enter their email address.\n3. Choose a role — **Member** or **Admin**. For an admin, also choose what they may do; each\n   capability is explained in [Roles](/help/hub/roles/).\n4. Click **Send Invite**.\n\n## What your teammate sees\n\nThey get an email with **Accept invitation** in it. The invitation is valid for **14 days**, and\nthey must sign in with the same email address you invited.\n\n**Heads up:** If they sign in with a different address — a personal Gmail instead of their work\nemail, say — it won't match the invitation. Invite the address they'll actually use.\n\nIf the invite has expired or got lost, sending it again refreshes it.\n\n## Good to know\n\n- **Seats limit invites.** Members plus pending invites count against your seats, so an invite\n  nobody has accepted still holds one. Add seats under [Subscriptions](/help/hub/subscription/).\n- **One organization at a time.** A person belongs to one organization at a time.\n\n## Removing someone\n\nWhen someone leaves, an owner can choose **Remove from Organization**:\n\n- they lose access immediately;\n- their seat is freed for someone else;\n- their usage history is kept, so your records stay complete.\n\nIf you invite them back later, they get the same account back.\n\n## Related\n\n- [Roles and what each can do](/help/hub/roles/) — decide between Member and Admin.\n- [Seats](/help/hub/seats/) — how many people your plan holds and how to add more.\n- [Team usage](/help/hub/team-usage/) — see how each teammate is using Crunch.\n","path":"/help/hub/inviting-users/","section":"Your account","slug":"hub/inviting-users","title":"Invite teammates","url":"https://bumbleb.co/help/hub/inviting-users/"},{"description":"Send a long analysis to the background, close the tab, and get told when it's ready.","gate":"background_turns","markdown":"\nA thorough question can keep Crunch working for several minutes — reading your data, checking what\nit found, and writing it up. You shouldn't have to keep a browser tab open and wait for it. With\nbackground turns, you send the analysis to the background, close the tab or move on to something\nelse, and Crunch tells you when it's ready. Say you ask on Friday afternoon for a week-by-week look\nat signups by channel: you can head off, and the answer is waiting, with an email saying so.\n\n## What you'll need\n\n- A Team plan (see [pricing](/pricing/)). Everyone on the plan gets background turns, including\n  teammates the plan covers.\n\nWithout it, nothing changes: closing the tab stops the analysis, as it always has.\n\n## Send an analysis to the background\n\n1. Ask your question as usual. While Crunch is working, a **Run in background** button appears next\n   to the **Stop generating** button in the message box.\n2. Click **Run in background**. It changes to **In background**, and a note says you can close the\n   tab and that you'll be notified when it's ready.\n3. Close the tab, or carry on elsewhere. Crunch keeps working.\n\nYou don't have to click the button first. If you close the tab while Crunch is working, your browser\nasks whether you want to leave the site; choose to leave and the analysis carries on in the\nbackground. And if you click **New chat** or open another conversation in the middle of an\nanalysis, Crunch asks what to do with it: **Run in background**, **Stop it**, or **Stay here**.\n\n## Find out it's ready\n\nWhen the analysis ends, you get a notification in the bell and an email, each with a link straight\nback to the conversation. It tells you one of three things:\n\n- **Your analysis is ready** — the answer is waiting.\n- **Your analysis needs your answer** — Crunch paused to ask you something, for example whether to\n  keep going after working for a while. Reply in the conversation and it carries on from there.\n- **Your analysis stopped before it finished** — something on our side cut it short. Open the\n  conversation to see what was done, and click **Retry** to run it again.\n\n## Come back while it's still working\n\nOpen the conversation any time. You'll see what Crunch has done so far, with the line **Working in\nthe background — this updates as it goes** under it, and the answer fills in as Crunch works. In\nyour conversation list, a small green dot marks a conversation that is still working.\n\n## Stop it\n\nClosing the tab never stops a background analysis; stopping it is always your choice. While you're\nwatching it, click **Stop generating** in the message box. If you've come back to the conversation\nlater, click **Stop** on the **Working in the background** line. The reply keeps what was done and\nsays **Stopped by you.** You aren't sent a notification for an analysis you stopped yourself.\n\n## How to tell it worked\n\nAfter you click **Run in background**, the button reads **In background**. Close the tab, and later\nthe bell shows **Your analysis is ready** (or **needs your answer**) with a link to the\nconversation, and the same arrives by email.\n\n## Good to know\n\n- **It counts as a running analysis.** Your plan allows a set number of analyses at once, and one in\n  the background is one of them. If you start another while it works, you may see \"You have too many\n  requests running at once. Wait for one to finish and try again.\" Stop the background one, or wait\n  for it, and ask again.\n- **A pause for your answer is a pause.** When Crunch stops to ask whether to keep going, it waits\n  for you rather than carrying on alone.\n- **The title may be generic.** If the analysis ends before the conversation has a title, the\n  notification calls it \"your question\".\n- **Follow-ups work as usual.** Ask a follow-up in the same conversation and Crunch builds on the\n  answer it gave in the background.\n\n## Related\n\n- [Notifications](/help/features/notifications/) — where your updates appear and how to read them.\n- [Ask Crunch about your data](/help/features/analyst/) — how an analysis works, step by step.\n- [Long conversations](/help/features/long-conversations/) — what happens as a conversation grows.\n","path":"/help/features/background-turns/","section":"Features","slug":"features/background-turns","title":"Keep an analysis running after you close the tab","url":"https://bumbleb.co/help/features/background-turns/"},{"description":"Ways to keep raw personal data — emails, phone numbers, customer records — from ever leaving your side.","gate":null,"markdown":"\nSome of the data that would make your questions sharper is also data you'd rather not hand to anyone:\ncustomers' emails, phone numbers, account records. You don't have to choose between the two. Each way\nof connecting data to Crunch has a way to keep the raw personal values on your side, so what Crunch\nreads is only what you've decided to let out.\n\n## On your website\n\nHash emails and phone numbers in the visitor's browser before they are sent. Hashing turns a value\nlike an email address into a scrambled code, so the address itself never travels. See\n[Hashing user IDs](/help/connect/hash-user-ids/) for how to set it up.\n\n## In your database (Team)\n\nThe [on-prem sync connector](/help/features/on-prem-sync/) runs in your network — on your own\nmachines — and applies a rule to every column before anything leaves. A connector is the piece that\nmoves data from one of your sources to Crunch. Each column's rule is one of:\n\n- **drop** — the column isn't sent at all;\n- **hash** — it's sent as a scrambled code instead of the raw value;\n- **round** or **truncate** — it's sent less precisely;\n- **send** — it's sent as it is.\n\nTwo safeguards sit underneath the rules:\n\n- **A column without a rule is dropped.** Forgetting to write a rule never lets a column out.\n- **A local floor file can force a column to be dropped or hashed whatever the rules say.** It's kept\n  locally, on your side.\n\n## In your data lake (Team)\n\n[Link tables](/help/features/external-connections/) with a read-only user — a login that can read but\nnot change anything — and only the schemas you allow. A schema is a named group of tables, so you\nchoose exactly which groups Crunch can see.\n\n## Need more?\n\nFor anything beyond these, talk to us about **Business**.\n\n## Related\n\n- [What we collect](/help/privacy/what-we-collect/) — exactly what each connection sends.\n- [Hashing user IDs](/help/connect/hash-user-ids/) — setting up hashing on your website.\n- [On-prem sync](/help/features/on-prem-sync/) — the rules and floor file in detail.\n","path":"/help/privacy/sensitive-data/","section":"Privacy \u0026 security","slug":"privacy/sensitive-data","title":"Keep sensitive data inside your network","url":"https://bumbleb.co/help/privacy/sensitive-data/"},{"description":"The context meter, and compacting a long chat so you can keep going.","gate":null,"markdown":"\nThe model behind Crunch can only read so much at once — call it its reading room. A long\nconversation, with many questions and answers, slowly fills that room. Crunch shows you how full it\nis, and when it gets full, it makes space so you can keep going in the same conversation.\n\n## Watch the meter\n\nThe pill beside the send button shows how much of the model's reading room the conversation uses\n(**Context used**). It turns amber from 70% and red from 90%.\n\n## Compacting\n\nWhen the room fills, Crunch **compacts** the conversation: from then on, the model reads a summary of\nthe earlier turns — your questions, its answers and the datasets they referred to — and the last\nturns in full.\n\n- It happens on its own at 70%.\n- To do it yourself, click **Compact now**, or type:\n\n```\n/compact\n```\n\nYour transcript on screen never changes. Compacting only changes what the model reads, not what you\nsee.\n\n## Good to know\n\n- **Tip:** If a message says the conversation has grown past what the model can read, type\n  `/compact` and ask your question again.\n\n## Related\n\n- [Ask Crunch about your data](/help/features/analyst/) — asking and reading answers.\n- [Choosing a model](/help/features/models/) — pick the model for your next message.\n","path":"/help/features/long-conversations/","section":"Features","slug":"features/long-conversations","title":"Long conversations","url":"https://bumbleb.co/help/features/long-conversations/"},{"description":"Updates from the BumbleB team, in the bell.","gate":null,"markdown":"\nThe bell in the chat header is where the BumbleB team tells you about updates. It's also where you\nhear back when a bug you reported is resolved, and where you're told that this month's included\nusage is nearly or fully used, so a busy week doesn't end in refused questions.\n\n## Read your updates\n\n1. Look for the bell in the chat header. A red badge means you have unread updates.\n2. Click an update to mark it read. Some carry a link, which opens in a new tab.\n\n## Good to know\n\n- Nothing is deleted. Older updates move down the list, so you can always scroll back to one.\n- Usage notices also arrive by email: one when you've used 90% of the month's included usage, and\n  one when it's used up. See [Credits](/help/hub/credits/) for what happens next.\n\n## Related\n\n- [Reporting a bug](/help/features/report-a-bug/) — tell us what went wrong; the reply arrives here.\n- [Keep an analysis running after you close the tab](/help/features/background-turns/) — get told here when a background analysis is ready.\n","path":"/help/features/notifications/","section":"Features","slug":"features/notifications","title":"Notifications","url":"https://bumbleb.co/help/features/notifications/"},{"description":"For the hard ones: Crunch explores your data, then proposes a plan you approve and run.","gate":"agents","markdown":"\nSome questions are really several questions. \"Why did revenue drop, and which channels should we\ncut?\" needs a handful of figures, each counted a particular way. The Plan agent lays out that work\nbefore any figure is computed, so you can see what will be answered, how, and from what data — and\nchange it — before anything runs.\n\n## What you'll need\n\n- The Plan agent, if your plan includes it.\n- The datasets your question is about, picked in the conversation — see\n  [Datasets](/help/features/datasets/).\n\n## Start it\n\n1. Click the agent pill in the message box. It shows **Analyst ▾**.\n2. Choose **Plan**. The choice sticks for the conversation.\n3. Ask your question as you normally would.\n\n## What you get\n\nA **plan** in the side panel, with one row per question to answer. Each row has:\n\n**You'll get** · **Counted as** · **From** · **Can't** · **Done when** · **Result**.\n\nWhile planning, Crunch only checks what the data holds — which columns there are, how fresh they\nare, and how filled in they are. It states no figures. That's on purpose: the plan is for agreeing\nthe work, not for answering yet.\n\nA single lookup gets answered as a lookup, not planned.\n\n## Change it\n\nEdit the plan in the panel and click **Save revision**. Your revision is what runs with your next\nmessage.\n\n## Run it\n\n1. Switch the pill back to **Analyst**.\n2. Say **go**.\n3. Crunch runs each row and writes its status and result back into the plan, as a new revision:\n   solved, partly, not in this data, or doubted.\n\n## How to tell it worked\n\nEach row in the plan shows its status and result: solved, partly, not in this data, or doubted.\n\n## Good to know\n\n- A big plan that reaches its time limit pauses. Say \"keep going\" to carry on.\n\n## Related\n\n- [Report agent](/help/features/report-agent/) — write up what the conversation settled.\n- [Ask Crunch about your data](/help/features/analyst/) — the default agent, for single questions.\n","path":"/help/features/plan-agent/","section":"Features","slug":"features/plan-agent","title":"Plan agent","url":"https://bumbleb.co/help/features/plan-agent/"},{"description":"The lead, the checks, how Crunch read you, how it counted, and what's still open.","gate":null,"markdown":"\nA figure is only useful if you know what it counts. That's why every Crunch answer leads with the\nresult and then shows its working: how it understood your question, which rules it counted by, what\nit couldn't confirm and what's still open. Knowing where each of those sits lets you trust a figure\nquickly — or spot in seconds that it answered a slightly different question from yours.\n\n## The parts of an answer\n\nThe result comes first. After it, you'll see some or all of these:\n\n- **↗** — after a figure, this opens the table or chart behind it, so you can see the numbers it came\n  from.\n- **⚑ Check before relying on this** — what the data could not confirm. Treat anything listed here as\n  a lead to look into, not a settled fact.\n- **How I read your question** — when a word in your question could mean more than one thing, this\n  says which reading Crunch took. It's the first thing to check.\n- **How this was counted** — the definitions and rules it actually used, so you can see whether\n  they match how your business counts.\n- **Not established** — what it could not answer, why, and what would answer it.\n- **Where this sits** — what is still open and what to ask next.\n\n## A good way to read one\n\n1. **Read the result**, so you know what Crunch is claiming.\n2. **Check How I read your question.** If the reading is wrong, the rest of the answer is about a\n   different question.\n3. **Look at How this was counted** for any figure you plan to repeat to someone else.\n4. **Note any ⚑ Check before relying on this** and **Not established** items before you act on the\n   answer.\n5. **Use Where this sits** to pick your next question.\n\n## When the reading isn't what you meant\n\nSay so in one line and ask again — for example, \"by visits I mean sessions, not page views.\" See\n[Correcting Crunch](/help/getting-started/correcting-crunch/).\n\n## Related\n\n- [Correcting Crunch](/help/getting-started/correcting-crunch/) — how to put a misread question back\n  on track in one line.\n- [Analyst agent](/help/features/analyst/) — more on each part of an answer.\n- [Slow, empty or \"not established\" answers](/help/trouble/answer-problems/) — what to do with a\n  \"not established\" or a check you can't resolve.\n","path":"/help/getting-started/reading-an-answer/","section":"Getting started","slug":"getting-started/reading-an-answer","title":"Read an answer","url":"https://bumbleb.co/help/getting-started/reading-an-answer/"},{"description":"Read tables where they live — Postgres or an Iceberg REST catalog — with no copy.","gate":"connections","markdown":"\nIf your data already lives in a Postgres database or behind an Iceberg REST catalog (a common way\nfor data lakes to list their tables), you don't have to export it to ask Crunch about it. You link\nthe tables, and Crunch reads them where they are — nothing is copied. That keeps one copy of your\ndata, in the place your team already looks after.\n\n## What you'll need\n\n- A Team plan (see [pricing](/pricing/)).\n- Sign-in details for your Postgres database or Iceberg REST catalog. **Use a read-only user**, so\n  Crunch can read your tables but never change them.\n\n## Add a connection\n\n1. Open the **Dataset Explorer** (the database icon in the chat header) and find **Connections**.\n2. Click **+ Add connection**, give it a name and choose the kind:\n   - **Postgres** — host, port, database, user, password, and optionally the schemas it may read.\n   - **Iceberg REST** — catalog URI, warehouse, and how it signs in (OAuth client, bearer token or\n     none). Optionally add the namespaces it may read, and a storage key if your catalog doesn't\n     vend storage credentials (hand out the keys to the files itself).\n3. Click **Add connection**.\n\n## Link a table\n\n1. Click **Link dataset**, choose the connection and enter the table: `schema.table`, or\n   `namespace.table` for Iceberg.\n2. Click **Read columns**.\n3. Name the dataset and tick the columns that make a row unique. If you tick none, every row is\n   treated as distinct. Set the time and user columns too, if the table has them.\n4. Click **Link dataset**.\n\n## How to tell it worked\n\nThe table appears in the datasets picker as a **linked source**, ready to pick in any conversation.\n\n## Good to know\n\n- A linked table can't be your clickstream, and it has no snapshots: Crunch reads it as it is now.\n- **Heads up:** Removing a connection removes its linked datasets too. Conversations that used them\n  can no longer read them.\n\n## Related\n\n- [Datasets](/help/features/datasets/) — pick a linked table for a conversation.\n- [On-prem sync connector](/help/features/on-prem-sync/) — for a database behind your firewall.\n- [Data home and lineage](/help/features/data-home/) — see where every dataset comes from.\n","path":"/help/features/external-connections/","section":"Features","slug":"features/external-connections","title":"Read your own data lake","url":"https://bumbleb.co/help/features/external-connections/"},{"description":"Tell us what went wrong, straight from the chat.","gate":null,"markdown":"\nIf something goes wrong — an error, a screen that won't load, an answer that looks broken — you can\ntell us without leaving the conversation. A clear report helps us find and fix it faster.\n\n## Send a report\n\n1. Click the bug icon (**Report a bug**) in the chat header.\n2. Describe what you did, what you expected, and what happened instead.\n3. In a saved conversation, tick **Allow the BumbleB team to view this conversation** if it would\n   help us. Without it, we see your description, your account, the conversation ID and basic browser\n   details, but not your messages.\n4. Click **Send**.\n\n## How to tell it worked\n\nWe reply through [Notifications](/help/features/notifications/) — the bell in the chat header.\n\n## Related\n\n- [Notifications](/help/features/notifications/) — where our reply arrives.\n- [Contact support](/help/trouble/contact-support/) — other ways to reach us.\n","path":"/help/features/report-a-bug/","section":"Features","slug":"features/report-a-bug","title":"Report a bug","url":"https://bumbleb.co/help/features/report-a-bug/"},{"description":"Write up what a conversation settled, as a document you can download.","gate":"agents","markdown":"\nAfter a good conversation, you often need to share what you found with someone who wasn't in it. The\nReport agent writes that up for you: one document, built from the answers the conversation already\nsettled, with its charts and tables in place.\n\n## What you'll need\n\n- The Report agent, if your plan includes it.\n- A conversation with some answers in it.\n\n## Start it\n\n1. Click the agent pill in the message box.\n2. Choose **Report**.\n\n## What you get\n\nA **Conversation report** in the side panel. It is built from the answers this conversation already\nsettled, with the charts and tables earlier turns drew.\n\nTo keep it or share it, click **Download PDF**.\n\n## What it won't do\n\n- It does not query your data again. Every figure in the report comes from an earlier answer, and\n  the report says so.\n- If nothing has been analysed yet, it asks the Analyst to compute the findings first, and says that\n  it did.\n\n## Related\n\n- [Plan agent](/help/features/plan-agent/) — lay out a big question before it runs.\n- [Charts, tables and artifacts](/help/features/artifacts/) — copy or download a single chart.\n","path":"/help/features/report-agent/","section":"Features","slug":"features/report-agent","title":"Report agent","url":"https://bumbleb.co/help/features/report-agent/"},{"description":"Business plans, shaped around your data, your security needs and your team.","gate":null,"markdown":"\nSome teams need more than a standard plan: more people, their own security or data requirements, or\njust someone to talk to directly. That's what the **Business** plan is for, and it starts with a\nconversation rather than a checkout.\n\n## When it's the right choice\n\nRequest a demo if you need:\n\n- **more than 10 seats** — a seat is one person on your plan;\n- **your own security or data needs** met;\n- **a direct line to us**.\n\nFor plan details, see [pricing](/pricing/).\n\n## Step by step\n\n1. Choose **Business → Request a demo** on the plan list, or go straight to\n   **hub.bumbleb.co/demo**.\n2. Tell us as much as you like about your team and what you need.\n\nWe'll get in touch.\n\n**Tip:** The more you tell us up front — how many people, what data you work with, any security\nrequirements — the more useful that first conversation will be.\n\n## Related\n\n- [Signing up and choosing a plan](/help/hub/signing-up/) — the other plans, and how to start.\n- [Seats](/help/hub/seats/) — how seats work on Team plans.\n","path":"/help/hub/request-a-demo/","section":"Your account","slug":"hub/request-a-demo","title":"Request a demo","url":"https://bumbleb.co/help/hub/request-a-demo/"},{"description":"Owner, admin and member: who can use Crunch, who runs the account, and who can change that.","gate":null,"markdown":"\nNot everyone on your team needs to see the bill. Roles let you give people exactly the access their\njob needs: an analyst can use Crunch without being able to cancel the subscription, while\nyour operations lead can manage seats and invoices. There are three roles — **owner**, **admin**\nand **member**.\n\n## Who can do what\n\n| Can… | Owner | Admin (by default) | Member |\n|---|---|---|---|\n| Use Crunch | ✓ | ✓ | ✓ |\n| Open the hub: Dashboard, your Profile and your Usage | ✓ | ✓ | ✓ |\n| See billing, credits and business details | ✓ | ✓ | — |\n| Change plan, seats, cancel | ✓ | ✓ | — |\n| Invite and manage users | ✓ | ✓ | — |\n| See the Organization page | ✓ | ✓ | — |\n| See seats | ✓ | ✓ | — |\n| Set up connectors and domains | ✓ | ✓ | — |\n| Set up a teammate's Crunch workspace | ✓ | ✓ | — |\n| See each teammate's usage by name | ✓ | ✓ | — |\n| Change someone's capabilities, remove members | ✓ | — | — |\n\nIn short:\n\n- **Members** use Crunch, and can open the hub to see their own profile and usage.\n- **Admins** use Crunch and, by default, help run the account.\n- **The owner** can do everything, including deciding what each admin can do and removing members.\n\n## The owner\n\nThe person who signs up is the owner.\n\n## Adjusting what an admin can do\n\nThe \"(by default)\" in the table matters. An owner can turn an admin's capabilities — the individual\npermissions in the table — on or off one by one. For example, an admin could manage users without\nbeing able to change the plan.\n\nNobody can change their own capabilities.\n\n**Tip:** You can see your own role and what it lets you do on your\n[Profile](/help/hub/profile/).\n\n## Related\n\n- [Invite teammates](/help/hub/inviting-users/) — choose a role when you invite someone.\n- [Your profile](/help/hub/profile/) — check your own role and capabilities.\n- [Team usage](/help/hub/team-usage/) — the by-name usage view in the table.\n","path":"/help/hub/roles/","section":"Your account","slug":"hub/roles","title":"Roles and what each can do","url":"https://bumbleb.co/help/hub/roles/"},{"description":"A seat is one person on a Team plan: how seats are assigned, the limits, and adding or removing them.","gate":"team","markdown":"\nOn a Team plan you pay per person, and each person's place on the plan is called a **seat**. Knowing\nhow seats work helps you plan ahead — for example, adding seats before a new hire's first day so\ntheir invitation can go out. Seats are a Team plan idea; see [pricing](/pricing/).\n\n## How seats are assigned\n\nEvery member of your organization — the owner included — holds one seat. You don't hand seats out\nyourself: a seat is assigned automatically when someone joins.\n\n**Seats** in the hub shows how many seats are:\n\n- **bought** — on your plan;\n- **held** — taken by people in your organization;\n- **free** — available for someone new.\n\n## How many seats you can have\n\n- Team plans have **2 to 10 seats**. More than 10 is the Business plan — see\n  [Request a demo](/help/hub/request-a-demo/).\n- Until your [business details](/help/hub/business-details/) are approved, an account is limited to\n  **3** seats. The seat picker — at sign-up and under **Subscriptions** — stops at 3 and says so,\n  and it opens up to your plan's full range once we approve your details.\n\n## Add or remove seats\n\n1. Open **Subscriptions → Change Plan / Seats**.\n2. Choose the number of seats you want.\n\nWhen you reduce seats:\n\n- **Fewer seats take effect at the end of the billing cycle.**\n- **You can't go below the number of people in your organization.** To drop seats, remove people\n  first.\n\n[Upgrading, downgrading and cancelling](/help/hub/subscription/) walks through the whole change,\nincluding when each kind is charged.\n\n## Good to know\n\nEach seat comes with its own monthly included usage — the amount of Crunch use your plan covers\neach month before [credits](/help/hub/credits/) are spent.\n\n## Related\n\n- [Invite teammates](/help/hub/inviting-users/) — fill a free seat.\n- [Business details](/help/hub/business-details/) — lift the 3-seat limit.\n- [Upgrading, downgrading and cancelling](/help/hub/subscription/) — the full seat-change steps.\n","path":"/help/hub/seats/","section":"Your account","slug":"hub/seats","title":"Seats","url":"https://bumbleb.co/help/hub/seats/"},{"description":"Put your ad spend next to what visitors do on your site, and see which spend comes back.","gate":null,"markdown":"\nYour ad platform says Meta's clicks are cheaper. But a cheap click that never signs up isn't a\nbargain. Cost per click tells you what a visit costs; what you really want to know is what a\n**sign-up** costs, campaign by campaign. That takes two datasets side by side: what you spent, and\nwhat visitors did once they reached your site. Crunch puts them together and answers in one go.\n\n## What you'll need\n\n- **Your site's events** — the record of what each visitor did, sent by the BumbleB tag in Google\n  Tag Manager (on a Shopify store, the BumbleB pixel instead).\n- **Your ad spend**, as an export from Google Ads, Meta or any other platform. Direct ad-platform\n  connectors aren't available yet, and you don't need one: the export is all it takes.\n- **Campaign tags on your ad links**, so a visit can be tied back to the campaign that brought it.\n\n## Step by step\n\n1. **Add the tag to your site.** Follow [Google Tag Manager](/help/connect/google-tag-manager/).\n   Make sure a sign-up sends its own event — Crunch can only count what your site sends.\n2. **Keep `utm_campaign` and `utm_source` on your ad links.** Most ad platforms add these for you.\n   They're what tells Crunch which campaign a visitor came from.\n3. **Export your spend by day and campaign**, and upload the file — see\n   [Ad spend without a connector](/help/connect/ad-spend/).\n4. **Name the campaign column the same way in both.** If the export calls it `campaign` and the\n   site calls it `utm_campaign`, rename one before uploading, so Crunch can line them up.\n5. **Pick both datasets** in the **datasets** pill, tell Crunch what it can't see (below), and ask.\n\n## Tell Crunch what it can't see\n\nYour spend export and your events don't know about each other, or about how your team talks. Say\nonce, at the start of the conversation:\n\n- **Which event is a sign-up** (or a lead, or a purchase) — \"a sign-up is a form submit on the\n  listings form\".\n- **How your campaigns are named** — \"listings-search is our Google campaign, listings-carousel is\n  Meta\".\n- **Untagged traffic** — \"a visit with no campaign is organic\".\n- **Currency**, if it isn't obvious from the export.\n\n## How to tell it worked\n\nBefore you ask anything, check the connector card in the hub: it shows **Active** once your domain\nis verified, and each API key's **Last used** date moves when events are getting through. Then, in\nthe chat, your answer should show spend and sign-ups for the same campaigns side by side. If a\ncampaign shows spend but no sign-ups, check it's spelled the same way in both places.\n\n## An example\n\nTwo campaigns, the same spend, and a suspicion that the cheap clicks aren't the good ones.\n\n| You ask | Crunch finds |\n|---|---|\n| We spent the same on our Google and Meta campaigns. How many sign-ups did each bring in, and what did each sign-up cost? | $12,000 on each. Google brought 960 sign-ups at $12.50 each; Meta brought 480 at $25.00 each — Google is about half the cost per sign-up. |\n\nSame budget, twice the sign-ups. That's the number to move budget on, and it's one neither the ad\nplatform nor the site could have given you alone.\n\n## Good to know\n\n- **No live ad-platform connectors yet.** Upload a fresh export when you want newer spend, and ask\n  about it in a new conversation.\n- **Spend and visits join on the campaign name.** A campaign named differently in the two places\n  won't line up until you rename it or tell Crunch they're the same.\n- **Events start on the day your tag goes live.** There's no history before that, so compare\n  spend only for dates your site was sending events.\n\n**Heads up:** your site only sends events once your domain is verified. Set that up first — see\n[Verify your domain](/help/hub/domains/).\n\n## Related\n\n- [Ad spend without a connector](/help/connect/ad-spend/) — export and upload your spend.\n- [Google Tag Manager](/help/connect/google-tag-manager/) — send your site's events.\n- [Ask one question across two datasets](/help/use-cases/two-datasets/) — how datasets line up.\n","path":"/help/use-cases/website-ad-roi/","section":"Use-case guides","slug":"use-cases/website-ad-roi","title":"See which ad spend comes back","url":"https://bumbleb.co/help/use-cases/website-ad-roi/"},{"description":"From the plan list to your first question in Crunch.","gate":null,"markdown":"\nSigning up takes a few minutes: pick a plan, confirm your email, tell us a little about yourself,\nand you're asking Crunch your first question. If you're not sure which plan fits yet, the Free plan\nneeds no card, so you can start there and change later.\n\n## What you'll need\n\n- An email address you can open right away (or a Google account).\n- Your country — it decides the currency you're billed in.\n- For a paid plan: a card, your billing contact and address, and, for Team, how many people will\n  use it.\n\n## Step by step\n\n1. **Pick a plan.** Go to **hub.bumbleb.co/signup**. You'll see three kinds of plan:\n   - **Starter** plans are one seat each — a seat is one person on the plan.\n   - **Team** plans are priced per seat, from two seats.\n   - **Business** is custom — click **Request a demo**.\n\n   The USD / INR toggle only changes the prices shown, not what you'll be billed in. Plan details\n   are on [pricing](/pricing/).\n2. **Confirm your email.** Enter your email and click **Send Magic Link** — a one-click sign-in link\n   sent to your inbox, valid for 24 hours. Or continue with Google instead.\n3. **Tell us about you.** Enter your name, an optional organization name, and your country. Your\n   country sets the currency you're billed in: India is billed in rupees; everywhere else in US\n   dollars.\n4. **Start your plan.**\n   - **Free plan:** confirm your details and click **Start Free**. No card needed.\n   - **Paid plan:** enter your billing contact and address (and, for Team, the number of seats),\n     then pay with Razorpay, the payment provider Crunch uses.\n5. **Open Crunch.** Click **Open Crunch** and ask your first question.\n\n## Trying Crunch first\n\nThere's no trial on paid plans: a paid plan is charged when you start it. To try Crunch first,\nstart on the **Free plan** — it comes with a limited amount of use every month, and you can move\nto a paid plan whenever you're ready.\n\n## How to tell it worked\n\nOnce you click **Open Crunch**, you're ready to ask your first question. Every new account also gets a one-time starter\ncredit — see [Credits](/help/hub/credits/) for what credits are.\n\n## Good to know\n\n- **The currency toggle is just for browsing.** What you're billed in comes from the country you\n  enter in step 3.\n- **On a Team plan, the seat picker stops at 3** until we approve your business details — see\n  [Seats](/help/hub/seats/).\n\n## Related\n\n- [Sign up and start on the free plan](/help/getting-started/sign-up/) — the shortest path in.\n- [Tour of the hub](/help/hub/tour/) — what each page in the hub is for.\n- [Invite teammates](/help/hub/inviting-users/) — bring your team in on a Team plan.\n- [Request a demo](/help/hub/request-a-demo/) — for more than 10 seats.\n","path":"/help/hub/signing-up/","section":"Your account","slug":"hub/signing-up","title":"Sign up and choose a plan","url":"https://bumbleb.co/help/hub/signing-up/"},{"description":"Start free with no card, connect a source, and ask your first question.","gate":null,"markdown":"\nThe quickest way to find out whether Crunch can answer your questions is to try it on your own data.\nThe **Free** plan lets you do that: it needs no card and no sales call. You sign up, connect a source\nor upload a file, and ask real questions before you pay anything.\n\n## What the Free plan includes\n\nIt includes a monthly amount of usage, one seat (one person who can sign in) and one conversation at\na time. For how it compares with the other plans, see [Plan levels](/pricing/).\n\n## Step by step\n\n1. **[Sign up](/help/hub/signing-up/)** at hub.bumbleb.co and choose **Free**. The hub is where your\n   account is run, and Crunch opens from it.\n2. Click **Open Crunch** to go from the hub into Crunch, where you ask questions.\n3. **Give Crunch some data to read.** Either [upload a file](/help/connect/upload-files/) you already\n   export, or [add the website tag](/help/connect/google-tag-manager/) so Crunch can read what\n   visitors do on your site.\n4. **Ask your first question** in your own words. For how to get to a good answer one step at a\n   time, see [Solve a problem in a conversation](/help/getting-started/solve-a-problem/).\n\n## Related\n\n- [Tour of the hub](/help/hub/tour/) — what each part of the hub is for.\n- [What Crunch is, and what it won't do](/help/getting-started/what-is-crunch/) — a short overview\n  before you start.\n- [Solve a problem in a conversation](/help/getting-started/solve-a-problem/) — how to ask so each\n  answer sets up the next.\n- [Plan levels](/pricing/) — what each plan includes.\n","path":"/help/getting-started/sign-up/","section":"Getting started","slug":"getting-started/sign-up","title":"Sign up and start on the free plan","url":"https://bumbleb.co/help/getting-started/sign-up/"},{"description":"What to do when an answer pauses, comes back incomplete, or isn't what you need.","gate":null,"markdown":"\nWhen an answer doesn't give you what you need — it pauses, says part of your question is \"not\nestablished\", or answers something slightly different from what you asked — there is almost always a\none-line fix. Find what you're seeing below.\n\n## It paused and asked to keep going\n\nThe question reached its time limit. Reply \"keep going\", or raise the limit under the brain button —\nsee [Analyst agent](/help/features/analyst/).\n\n## It says \"Not established\"\n\nYour data couldn't answer that part, and the answer says why and what would. Reply \"yes\" to have it\ntry, or supply the missing piece yourself.\n\n## It read your question differently\n\nCorrect it in one line and ask again — for example, \"by visits I mean sessions, not page views.\" See\n[Correcting Crunch](/help/getting-started/correcting-crunch/).\n\n## It says \"Check before relying on this\"\n\nA value didn't hold up when it was checked. Treat it as a lead to look into, not a result.\n\n## The conversation grew too long\n\nType this and ask again:\n\n```\n/compact\n```\n\nFrom then on Crunch reads a summary of the earlier turns, so there's room to carry on. See\n[Long conversations](/help/features/long-conversations/).\n\n## Your new upload isn't in the answer\n\nA conversation keeps the data it started with, so a file you uploaded after it began isn't part of it.\nAsk in a new conversation.\n\n## Still stuck?\n\n[Report a bug](/help/features/report-a-bug/) from the chat.\n\n## Related\n\n- [Reading an answer](/help/getting-started/reading-an-answer/) — what each part of an answer means.\n- [Contact support](/help/trouble/contact-support/) — other ways to reach us.\n","path":"/help/trouble/answer-problems/","section":"Trouble \u0026 contact","slug":"trouble/answer-problems","title":"Slow, empty or \"not established\" answers","url":"https://bumbleb.co/help/trouble/answer-problems/"},{"description":"Get to the answer one small question at a time — the way an analyst works.","gate":null,"markdown":"\nSomething feels off — enquiries seem weak, a campaign isn't paying back, sales dipped last month —\nand you want to know why. You know your business and what you're after; Crunch knows how to count.\nThe fastest way to an answer is to go one small step at a time, in your own words, and let each\nanswer set up the next question. That's how a good analyst works, and it keeps every answer short\nenough to check.\n\n## What you'll need\n\nSome data for Crunch to read — an uploaded file, your website or your store. If you haven't set that\nup yet, start with [Sign up and start on the free plan](/help/getting-started/sign-up/).\n\n## The five steps\n\n1. **Trend — what moved overall?** Start with the big picture: one number over time, such as\n   enquiries or orders each month. This tells you whether something really changed, and when.\n2. **Locate — where did it move?** Split that number by something that matters to your business —\n   city, product, channel — to find the part that behaves differently from the rest.\n3. **Drill — which part exactly, and when?** Narrow into that part until you can name it precisely,\n   and pin down when the change began.\n4. **Check — what rules each explanation in or out?** By now you'll have a few possible explanations.\n   Ask for the one fact that would tell them apart.\n5. **Conclude — which explanation fits the data better?** Put your explanations side by side and ask\n   which one the data supports.\n\nBy the end, a vague worry has a name: a specific part of the business, a specific time, and the\nexplanation that fits.\n\n## How to ask\n\n- **Ask one thing per turn.** A single question gets a single answer you can check before you build\n  on it.\n- **Use business words** — \"enquiries\", \"Android\", \"Pune\" — not table or column names. Crunch works\n  out which data holds them.\n\n## Say what you already know, once\n\nIf a value means something special, or you know how two things map, say it at the start: \"a rating of\n0 means not rated\", \"country code 1 is India, in rupees\". Crunch keeps it for the rest of the\nconversation, so you don't need to repeat it.\n\n## When Crunch reads you wrong\n\nEvery answer starts by saying how Crunch read your question. If that isn't what you meant, correct it\nin one line — \"use the app the enquiry came from, and include web\" — and ask again. See\n[Correcting Crunch](/help/getting-started/correcting-crunch/).\n\n## Related\n\n- [Reading an answer](/help/getting-started/reading-an-answer/) — what each part of an answer tells\n  you.\n- [Correcting Crunch](/help/getting-started/correcting-crunch/) — how to put a misread question back\n  on track.\n- [Slow, empty or \"not established\" answers](/help/trouble/answer-problems/) — what to do when an\n  answer isn't what you need.\n","path":"/help/getting-started/solve-a-problem/","section":"Getting started","slug":"getting-started/solve-a-problem","title":"Solve a problem in a conversation","url":"https://bumbleb.co/help/getting-started/solve-a-problem/"},{"description":"Ask in words, get SQL, and run it over your picked datasets a page at a time.","gate":"tsql","markdown":"\nSometimes you want the rows themselves, not a written answer — a list to check, a table to export, a\nquery you can tweak. The SQL console turns a question in plain words into SQL (the language\ndatabases are queried in), tells you how it understood you, and lets you edit and run the query over\nthe datasets picked in your conversation. You don't need to write SQL yourself, though you can.\n\n## What you'll need\n\n- The SQL console, if your plan includes it.\n- Datasets picked in the conversation — see [Datasets](/help/features/datasets/).\n\n## Open it\n\nClick the terminal icon in the chat header — the console opens, titled **tsql**.\n\n## Use it\n\n1. The **Datasets** chips are the datasets picked in this conversation. Click one to insert it into\n   your query; `{{clickstream}}` is your event stream. To pick datasets, use the controls under the\n   message box.\n2. Type a question — for example \"orders per week by channel, with the refund share\" — and click\n   **Draft SQL**. The first question on a new pick can take a minute.\n3. Read **What I understood**: the reading it took, what is counted, its assumptions, anything **Not\n   in your data**, and what to **Check before trusting**.\n4. Edit the SQL if you like, **Check** it, then **Run**.\n5. Results come 100 rows a page. **Count rows** gives the total.\n\n## How to tell it worked\n\nRows appear under the query, a page at a time, and **Count rows** tells you how many there are in\nall.\n\n## Good to know\n\n- A query is one read-only `SELECT` (a `WITH` is allowed), over the picked datasets only, joining up\n  to 8 of them. Read-only means it can look at your data but never change it.\n- Queries that would be too expensive are refused, and the console tells you why.\n- Paging stops at row 10,000. If you need more, narrow the query or aggregate it (count or sum by\n  group) instead.\n- A query reads each dataset at the same version the chat reads, so the console and your answers\n  agree.\n\n## Related\n\n- [Data home and lineage](/help/features/data-home/) — run SQL on any dataset and keep your queries.\n- [Datasets](/help/features/datasets/) — pick what the console can read.\n- [Ask Crunch about your data](/help/features/analyst/) — get a written answer instead.\n","path":"/help/features/sql-console/","section":"Features","slug":"features/sql-console","title":"SQL console","url":"https://bumbleb.co/help/features/sql-console/"},{"description":"Two editions of the analyst. Both count correctly; Pro works harder on the hard ones.","gate":"eyela_pro","markdown":"\nThe analyst comes in two editions, Starter and Pro. Both get the same figures right — correctness is\nin both editions, so a Starter answer is not a rougher draft of a Pro one. What **Pro** adds is\ndepth: it checks more, explains more tightly, and works through harder questions.\n\n## What you'll need\n\n- Nothing to switch on: your plan decides the edition. Team plans get Pro, and Starter plans get\n  Starter (see [pricing](/pricing/)). If you're on a teammate's Team plan, you get Pro too.\n\n## What Pro adds\n\n- **Works to a named method.** Each question is matched to a known kind of job and worked step by\n  step, so the same kind of question gets the same method every time. That makes answers to similar\n  questions easy to compare.\n- **Tests its own doubts.** When a figure looks off, Pro checks it against the rows — does it agree,\n  sum to the whole, survive a recount? — and tells you only about doubts that failed a check, so you\n  aren't warned about things that turned out fine.\n- **Reads your words one way, and says so.** Every filter traces back to something you asked. An\n  ambiguous word gets one reading, with the other given once.\n- **Handles messy joins.** When two tables are matched on a shared column (a join key) and some rows\n  are missing that key, those rows are counted apart and stated, not silently dropped.\n- **Tighter answers.** A ranking names its leaders, a definition is said once, and the next question\n  follows from the figures shown.\n- **Says when a window is outside your data**, and owns each open part with the exact next step.\n\n## How to tell which you have\n\nLook beside your name: a **Pro** badge means your account is on Pro.\n\n## Related\n\n- [Ask Crunch about your data](/help/features/analyst/) — how to read an answer in either edition.\n- [Pricing](/pricing/) — which plans include which edition.\n","path":"/help/features/starter-vs-pro/","section":"Features","slug":"features/starter-vs-pro","title":"Starter and Pro","url":"https://bumbleb.co/help/features/starter-vs-pro/"},{"description":"Starters for your data, and ideas for what to ask next.","gate":null,"markdown":"\nA blank message box can be the hardest part. Crunch suggests questions you can ask of the data you\npicked, and after each answer it offers a few follow-ups, so you always have a next step to click.\n\n## Before you start\n\nOnce you've picked datasets, up to four cards suggest things you can find out from them. Click one\nto ask it.\n\n## After an answer\n\n**Ask next** offers up to three follow-ups. They never repeat what you asked or what was offered\nbefore. Clicking one sends it.\n\n## Good to know\n\n- Suggestions cost nothing until you send one.\n\n## Related\n\n- [Datasets](/help/features/datasets/) — pick the data the starter cards are drawn from.\n- [Ask Crunch about your data](/help/features/analyst/) — asking your own question.\n","path":"/help/features/next-questions/","section":"Features","slug":"features/next-questions","title":"Suggested questions","url":"https://bumbleb.co/help/features/next-questions/"},{"description":"A container in your network anonymizes every field by your rules before anything syncs.","gate":"bb_connector","markdown":"\nSome data can't leave your network as it is — customer records, for example. The on-prem sync\nconnector lets you analyse it anyway. You run a small container (a packaged program) inside your\nnetwork. It reads your database with a read-only user, applies your rules to every field before\nanything leaves, and pushes the result out over outbound HTTPS. You don't open any inbound port, and\nyou decide, column by column, exactly what Crunch ever sees.\n\nIt syncs **MongoDB** today.\n\n## What you'll need\n\n- A Team plan (see [pricing](/pricing/)).\n- A machine in your network that can run Docker and reach your database.\n- A read-only user for the database.\n\n## Set it up\n\n1. In the hub, open **Connectors**, click **+ Create connector** and choose **On-prem\n   database**. Give it a name.\n2. **Install the container.** Click **Get an enrollment token** — it is shown once and is valid for\n   24 hours, so copy it now. Then run the `docker run` command on the card, with your database's\n   address and name.\n3. **Approve the machine.** When the container starts, it appears on the card with a fingerprint (a\n   short code that identifies that machine). Click **Approve** only if the fingerprint matches the\n   one in the container's log — that's how you know it's your container and not someone else's.\n4. **Decide what leaves.** For each table, set the collection, how it syncs (merge updates, append\n   new, or full refresh), its key and its changed-at field — and a rule for **every** column.\n5. **Accept the rules on your side.** Rules that send more data wait until someone runs this on the\n   container:\n\n```\nbb-connector accept-rules\n```\n\nThat last step means a change that sends more data out always needs someone inside your network to\nagree to it.\n\n## Rules\n\n| Rule | What leaves |\n|---|---|\n| Drop | nothing |\n| Hash with your key | a hash only your key can make |\n| Hash as email / Hash as phone | a hash that matches the website tag's [hashed user IDs](/help/connect/hash-user-ids/) |\n| Round time | to the hour, day or month |\n| Keep first 3 / last 4 characters | part of the value |\n| Keep email domain only | `example.com` |\n| Send as is | the value |\n\nA hash turns a value into a scrambled code: the same value always gives the same code, so records\ncan still be matched, but the original can't be read back from it.\n\n**Heads up:** A column without a rule is dropped.\n\n## Join with your website data\n\nTo match these records with your website's visitors, use **Hash as email/phone** on both sides, or\n**Send as is** for an internal ID.\n\n**Heads up:** A value hashed with your key never matches your website's events, so don't use **Hash\nwith your key** on a column you want to join on.\n\n## How to tell it worked\n\n**What landed** shows each run — table, mode, status and rows. From there you can **Pause**,\n**Resume** and **Resync**. It syncs every hour by default.\n\n## Related\n\n- [Hash user IDs](/help/connect/hash-user-ids/) — how the website tag hashes emails and phones.\n- [Your own data lake](/help/features/external-connections/) — read Postgres or Iceberg tables where\n  they live.\n","path":"/help/features/on-prem-sync/","section":"Features","slug":"features/on-prem-sync","title":"Sync a database from behind your firewall","url":"https://bumbleb.co/help/features/on-prem-sync/"},{"description":"Teammates, seats and team usage on a Team plan.","gate":"team","markdown":"\nWhen more than one person in your company uses Crunch, a Team plan lets you bring them in under one\naccount and see who is using what. A Team plan is priced per seat, from two seats, and every seat\ncomes with its own monthly usage.\n\n## What you'll need\n\n- A Team plan (see [pricing](/pricing/)).\n\n## If you're on a teammate's Team plan\n\nWhen a colleague invites you onto their Team plan, you don't need a plan or a card of your own.\nYour seat is covered by the person who pays for the plan, usually the owner.\n\n- **You'll see whose plan it is.** Your hub dashboard says \"Your seat is on *their name*'s *plan\n  name* plan — they manage the subscription and billing.\"\n- **Your usage is your own.** **Usage** in the hub shows what you've used this month against your\n  seat's allowance, the usage that comes with each seat.\n- **They pay for it.** If you go past your allowance, the extra is charged to the payer's credits\n  (prepaid usage on their account), not to you.\n- **Billing is theirs.** **Subscriptions**, **Billing** and **Credits** belong to the payer, so you\n  won't see them in your sidebar.\n- **Your access follows their plan.** If their plan lapses, Crunch stops for you too until it's\n  back. Ask the payer, or whoever manages your team, to sort it out.\n\n## Where to go next\n\n- [Inviting users](/help/hub/inviting-users/) — add teammates as members or admins.\n- [Roles](/help/hub/roles/) — who can do what.\n- [Seats](/help/hub/seats/) — one per person, 2 to 10.\n- [Team usage](/help/hub/team-usage/) — what each teammate used.\n","path":"/help/features/team-management/","section":"Features","slug":"features/team-management","title":"Team management","url":"https://bumbleb.co/help/features/team-management/"},{"description":"See how much each teammate uses Crunch, by name, over any dates you pick (Team plans).","gate":"team","markdown":"\nWhen the month's usage climbs faster than expected, the natural question is \"who's using it, and\non what?\" **Team Usage** answers that: it breaks your organization's usage down person by person,\nso you can spot who relies on Crunch most and plan seats and credits around it. It's for Team\nplans — see [pricing](/pricing/).\n\n## What you'll need\n\nBecause this page shows individual figures by name, only owners and people given *View Team Usage*\ncan see it. See [Roles](/help/hub/roles/).\n\n## Step by step\n\n1. In the hub, open **Team Usage**. It starts on this month.\n2. To look at another period, pick **From** and **To** dates.\n3. Click **Apply**.\n\n## What it shows\n\nFor every member:\n\n- their **role**;\n- the **models** they used — the AI models behind Crunch's answers;\n- their **activity**;\n- **tokens** — the units models measure text in;\n- **requests**;\n- **credits** spent.\n\nAn **organization total** sits alongside, so you can see each person against the whole.\n\n## Good to know\n\n**Each teammate has their own included usage.** On a Team plan, once a teammate's own included usage\nruns out, their use draws on the account's [credits](/help/hub/credits/). One heavy user can spend\nthe shared credits even while everyone else is well within their own allowance.\n\n## Related\n\n- [Your usage](/help/hub/usage/) — your own usage against your plan.\n- [Credits](/help/hub/credits/) — the shared balance teammates draw on.\n- [Choosing a model](/help/features/models/) — why some questions use more than others.\n","path":"/help/hub/team-usage/","section":"Your account","slug":"hub/team-usage","title":"Team usage","url":"https://bumbleb.co/help/hub/team-usage/"},{"description":"hub.bumbleb.co: where you run your account — plan, billing, credits, team, connectors and domains.","gate":null,"markdown":"\nCrunch has two parts. **Crunch** itself is the chat, where you ask questions about your data. The\n**hub**, at **hub.bumbleb.co**, is where the account is run: the plan you're on, what you've paid,\nwho's on your team, and the connectors that bring your data in. Crunch opens from the hub.\n\nIf you're setting things up for the first time, or someone handed you the account, this page tells\nyou what each part of the hub is for.\n\n## Who uses the hub\n\nEveryone on your account can open the hub. Members see the **Dashboard**, their **Profile** and\ntheir own **Usage** — the number they are measured against. Owners, and admins by default, also\nsee the pages that run the account: billing, the team, connectors and domains.\n\nYou see only the pages your role allows, so if a page below is missing from your sidebar, that's\nwhy. See [Roles](/help/hub/roles/).\n\n## What's in the sidebar\n\n| In the sidebar | What it is for |\n|---|---|\n| **Dashboard** | Your plan at a glance, with next steps |\n| **Profile** | Your account and role (read-only) |\n| **Organization** | Your organization's details (Team plans) |\n| **Subscriptions** | Change plan or seats, cancel, payment method |\n| **Billing** | Invoices and payments |\n| **Credits** | Your balance, buying credits, recent activity |\n| **Usage** | What you used this month against your plan |\n| **Team Usage** | Each teammate's usage (Team plans) |\n| **Business Details** | Details for your invoices |\n| **Users** | Invite and manage teammates |\n| **Seats** | How many seats your plan has and who holds them (Team plans) |\n| **Connectors** | Your website tag, Shopify pixel and on-prem sync |\n| **Domains** | Verify the sites your connectors collect from |\n\n## Where to start\n\nIf you've just signed up, these pages are worth a look first:\n\n- **Connectors** and **Domains** — Crunch can only answer questions about data it receives. See\n  [Connectors in the hub](/help/hub/connectors/) and [Verify your domain](/help/hub/domains/).\n- **Users** — on a Team plan, invite the people who'll ask questions. See\n  [Invite teammates](/help/hub/inviting-users/).\n- **Usage** and **Credits** — keep an eye on how much you're using. See\n  [Your usage](/help/hub/usage/) and [Credits](/help/hub/credits/).\n\n## Related\n\n- [Roles and what each can do](/help/hub/roles/) — why you see the pages you do.\n- [Sign up and choose a plan](/help/hub/signing-up/) — getting your account started.\n- [What Crunch is, and what it won't do](/help/getting-started/what-is-crunch/) — the chat side\n  of things.\n","path":"/help/hub/tour/","section":"Your account","slug":"hub/tour","title":"Tour of the hub","url":"https://bumbleb.co/help/hub/tour/"},{"description":"Clickstream analytics in conversation: funnels, sources, pages and who your heavy users are.","gate":null,"markdown":"\nYour traffic is up, but sign-ups aren't, and you'd like to know where people lose interest. These\nare the questions a product or growth team asks every week — where people drop, which source\nconverts, which pages hold attention, whether a number is broad use or a few heavy users. Crunch\nanswers them from your site's or app's events: the record of each thing a visitor did, like\nviewing a page or signing up. You ask in plain words; there are no reports to build first.\n\n## What you'll need\n\n- **Your events**, sent by the BumbleB tag in Google Tag Manager — or, on a Shopify store, the\n  BumbleB pixel.\n- **A verified domain**, so your site's events are accepted.\n- Optionally, **the business side** — revenue, CRM or ad spend — as uploads, so you can ask\n  across both.\n\n## Step by step\n\n1. **Add the tag.** Follow [Google Tag Manager](/help/connect/google-tag-manager/).\n2. **Send the events that matter to you** — sign-ups, key feature use, errors — not only page\n   views. Crunch can only answer about what you send, so a step you never track is a step it can't\n   see.\n3. **Hash user IDs if you identify users.** Hash them in the page before anything is sent — see\n   [Hashing user IDs](/help/connect/hash-user-ids/). That way you can follow people across visits\n   without sending who they are.\n4. **Add the business side when you need it.** See\n   [Which route for which data](/help/connect/which-route/).\n5. **Pick your events dataset** in the **datasets** pill, tell Crunch what it can't see (below),\n   and ask.\n\n## Tell Crunch what it can't see\n\nEvent names are short, and their meaning lives in your team's heads. Say once, at the start:\n\n- **Which event means what** — \"activation is a first project created\", \"scroll_75 means the page\n  was read all the way down\".\n- **Who counts as active** — a visit, a session, a signed-in user. Each gives a different number,\n  and none of them is wrong.\n- **Internal traffic** — \"ignore visits from our office domain\", so your own team doesn't inflate\n  the figures.\n\n## How to tell it worked\n\nOn the connector card in the hub, the status shows **Active** once your domain is verified, and\neach API key's **Last used** date moves as events arrive. In the chat, ask something simple first —\n\"how many visits did we get yesterday?\" — and check the figure against what you'd expect.\n\n## An example\n\nA listings site wants to know which of its pages actually earn their traffic.\n\n| You ask | Crunch finds |\n|---|---|\n| Which of our pages actually hold attention, and which just get a glance? Rank them by how often their visits get read all the way down. | Pricing 70% (5,600 of 8,000 visits), the neighborhood guide 45% (2,700 of 6,000), the blog post 20% (1,200 of 6,000). Pricing holds attention; the blog post mostly gets a glance. |\n\nThe neighborhood guide and the blog post drew the same 6,000 visits, but the guide kept more than\ntwice as many readers to the end. Raw traffic alone would have ranked them equal.\n\n## Good to know\n\n- **Crunch reads the events you send.** If a step isn't tracked, it says so instead of guessing.\n- **Events start on the day the tag goes live.** There is no history before it.\n- **Say the definition once.** \"Read all the way down\" or \"active\" can each be counted several\n  ways. Telling Crunch which one you mean at the start keeps every later answer on the same\n  footing.\n\n## Related\n\n- [Google Tag Manager](/help/connect/google-tag-manager/) — set up the events.\n- [See which ad spend comes back](/help/use-cases/website-ad-roi/) — put ad spend next to your\n  events.\n- [Solve a problem in a conversation](/help/getting-started/solve-a-problem/) — one small question\n  at a time.\n","path":"/help/use-cases/product-behavior/","section":"Use-case guides","slug":"use-cases/product-behavior","title":"Understand product use and behavior","url":"https://bumbleb.co/help/use-cases/product-behavior/"},{"description":"Change your plan or seats, cancel, and know when each change takes effect and what it costs.","gate":null,"markdown":"\nYour needs change: a new hire joins, a project winds down, or you outgrow the plan you started on.\n**Subscriptions** is where you change your plan or seats, or cancel. The main thing to understand\nis *timing* — some changes happen straight away and some wait for the end of your billing cycle\n(the period each payment covers) — and this page walks you through both.\n\n## What you'll need\n\nAccess to **Subscriptions**: owners, and admins with *Manage Subscriptions*. See\n[Roles](/help/hub/roles/).\n\n## Change your plan or seats\n\n1. Open **Subscriptions** and click **Change Plan / Seats**.\n2. Choose the plan and, for Team, the number of seats — a seat is one person on the plan.\n3. Choose when it happens: **Apply Now** or **At End of Billing Cycle**.\n4. Click **Preview Changes** and read the summary, so you know exactly what will change and what\n   you'll be charged.\n5. Click **Confirm Change** — or **Continue to Payment**, if there's something to pay.\n\n## When each change takes effect\n\n| Change | When | Charge |\n|---|---|---|\n| Upgrade, applied now | immediately | the difference for the days left in the cycle |\n| Downgrade or fewer seats | end of the billing cycle | none, no refund |\n| Moving to Free | end of the billing cycle | paid access continues until then |\n\n## How to tell it worked\n\nA scheduled change shows as a banner on **Subscriptions**. Changed your mind? Click\n**Cancel Change** on that banner.\n\n## Moving from Team to Starter\n\nMoving a team down to Starter takes one step first: your organization has to fit the new plan's\nseats. If it has more people than the new plan has seats, the change is turned down with\n\"Remove users first\". Remove your teammates on **Users** (see\n[Invite teammates](/help/hub/inviting-users/)), then make the change.\n\nWhen the move takes effect, the Team features end with the plan:\n\n- the Pro edition of the analyst — you get Starter (see [Starter and Pro](/help/features/starter-vs-pro/))\n- [background turns](/help/features/background-turns/), so closing the tab stops an analysis again\n- [Data home](/help/features/data-home/)\n- [your own data lake connections](/help/features/external-connections/)\n- [on-prem sync](/help/features/on-prem-sync/)\n\nWhich plan includes what is on [pricing](/pricing/).\n\n## Cancel your subscription\n\n1. Click **Cancel Subscription**.\n2. A paid subscription stays active until the end of the billing period. If you'd rather end it\n   immediately, choose **End Now**.\n\nTo come back later, click **Resubscribe**.\n\n**Tip:** If you only want to stop paying, move to the Free plan instead of cancelling. It keeps your\naccount and your data.\n\n## Good to know\n\n- **You can't drop below the number of people in your organization.** If you want fewer seats than\n  you have people, remove people first — see [Invite teammates](/help/hub/inviting-users/).\n- **Downgrades aren't refunded.** A downgrade or a cut in seats takes effect at the end of the\n  billing cycle, with no refund for the rest of it.\n\n## Related\n\n- [Seats](/help/hub/seats/) — how seats are counted and the limits on them.\n- [Billing and invoices](/help/hub/billing/) — see the charges a change produces, and update your\n  card.\n- [Credits](/help/hub/credits/) — buy more usage without changing your plan.\n","path":"/help/hub/subscription/","section":"Your account","slug":"hub/subscription","title":"Upgrade, downgrade or cancel","url":"https://bumbleb.co/help/hub/subscription/"},{"description":"Turn a file you already export into a dataset you can ask about.","gate":null,"markdown":"\nPlenty of what you'd like to ask about already lives in a file: an orders export, a spend report,\na list of listings. Upload it, and it becomes a dataset — a table in your own workspace that Crunch\ncan read and answer questions about. There's no tag or connector to set up.\n\n## What you'll need\n\n- A file with rows and columns, in `.csv`, `.tsv`, `.xlsx`, `.xls` or `.parquet`.\n- The right to upload it. You'll confirm this when you upload.\n\n## Step by step\n\n1. In a chat, click the **datasets** pill (the database icon) in the message box.\n2. Choose **Upload a file…**.\n3. Choose or drop the file, tick the box confirming you have the right to upload it, and click\n   **Upload**.\n4. Check the preview. It shows the first rows, each column's name and type, and any issues. For an\n   Excel file, pick the **Sheet** you want. If the column names aren't in the first row, set the\n   **Header row**.\n5. Choose **New dataset** and give it a name — lowercase letters, digits and `_`, starting with a\n   letter. Or, if this is a fresher copy of something you uploaded before, choose **New version\n   of** that upload.\n6. Click **Load**. You can close the dialog while it works; the upload keeps going.\n\n## How to tell it worked\n\nWhen loading is done, the dataset is selected for the conversation, and you can start asking.\n\n## Good to know\n\n- **Formats:** `.csv`, `.tsv`, `.xlsx`, `.xls` and `.parquet`, with one sheet per upload. A\n  workbook with several sheets means several uploads.\n- **Size:** up to **100 MB** per file, and up to **300 columns**.\n- **Values are kept as written.** Crunch doesn't guess, so:\n  - a date that could be day/month or month/day stays text — write dates as `YYYY-MM-DD` to avoid\n    this;\n  - numbers with leading zeros, like a ZIP code `02134`, stay text, so the zeros aren't lost;\n  - times without a time zone are read as UTC.\n\n## When you have newer numbers\n\nUpload the fresh export as **New version of** the same dataset. A conversation keeps reading the\ndata it started with, so ask about the new version in a new conversation.\n\n## Related\n\n- [Datasets: picking, pinning and versions](/help/features/datasets/) — choose what a\n  conversation reads.\n- [Any dataset](/help/use-cases/any-dataset/) — a worked example on one uploaded file.\n- [Fixes](/help/connect/fixes/) — when an upload is rejected.\n","path":"/help/connect/upload-files/","section":"Connect your data","slug":"connect/upload-files","title":"Upload an Excel, CSV or Parquet file","url":"https://bumbleb.co/help/connect/upload-files/"},{"description":"Ask your data questions from Claude Code, Claude Desktop or claude.ai.","gate":null,"markdown":"\nIf you already work in Claude, you can ask your data questions there instead of switching to\nCrunch. Crunch is a remote MCP server — MCP is the standard way Claude connects to outside tools —\nso once you add it, Claude can use Crunch to answer questions like \"how's activation trending?\".\n\nThe server URL is:\n\n```\nhttps://sequoia.bumbleb.co/mcp\n```\n\n## What you'll need\n\n- A Crunch account, and Claude Code, Claude Desktop or claude.ai.\n\n## Add it to claude.ai and Claude Desktop\n\n1. On claude.ai, open **Settings → Connectors** and choose **Add custom connector**.\n2. Name it `Sequoia` and enter the URL above.\n3. It appears under **+** in any conversation. Claude Desktop picks it up from your claude.ai account,\n   so there's nothing more to do there.\n\n## Add it to Claude Code\n\n1. Run this in your terminal:\n\n   ```bash\n   claude mcp add --transport http sequoia https://sequoia.bumbleb.co/mcp\n   ```\n\n   Add `--scope user` to have it in every project, or `--scope project` to share it with your team\n   through `.mcp.json`.\n2. If `claude mcp list` shows **Needs authentication**, run `/mcp` in a session, pick `sequoia`,\n   choose **Authenticate**, and approve in the browser.\n3. When you see **✔ Connected**, it's ready.\n\n## How to tell it worked\n\nJust ask: \"how's activation trending?\". Claude calls Crunch and answers from your data.\n\n**Tip:** To skip the permission prompt each time, allow `mcp__sequoia__*` in\n`.claude/settings.json`.\n\n## If it doesn't connect\n\n- Run `curl -I https://sequoia.bumbleb.co/mcp`. A 404 or 405 means the server is up; 401 or 403\n  means it wants you to sign in; nothing at all means a network problem or a typo.\n- Check for a stray space in a pasted URL.\n- If it's slow to start, run `MCP_TIMEOUT=60000 claude`.\n- **No MCP servers configured** means you are in a different project from the one where you added\n  it.\n\n## Related\n\n- [Ask Crunch about your data](/help/features/analyst/) — how Crunch answers and shows its counting.\n- [Datasets](/help/features/datasets/) — the data Crunch reads.\n","path":"/help/features/use-from-claude/","section":"Features","slug":"features/use-from-claude","title":"Use Crunch from Claude","url":"https://bumbleb.co/help/features/use-from-claude/"},{"description":"Prove the site a connector collects from is yours, by adding one DNS record.","gate":null,"markdown":"\nBefore Crunch accepts any data from your website or store, it needs to know the site is really\nyours. You prove that by adding one small record to your domain's settings — the same kind of check\nGoogle and most email tools use. Once it's done, events sent under your connector from anywhere\nelse are rejected — so data can only arrive under your connector from a site you control.\n\nUntil a domain verifies, a website or Shopify connector accepts no events at all — they're turned\naway rather than stored. It's worth doing this first.\n\n## What you'll need\n\n- A website or Shopify connector, created on the [Connectors](/help/hub/connectors/) page.\n- Access to your **domain registrar** — the company you bought your domain from, or wherever your\n  domain's DNS is managed. DNS is the internet's address book; a **DNS record** is one entry in it.\n\n## Step by step\n\n1. In the hub, open **Domains**.\n2. Under your connector, enter your root domain — for example `example.com` — and click\n   **Add Domain**. You don't need to add subdomains such as `www.` and `shop.`; they're covered\n   automatically.\n3. At your domain registrar, publish the record Crunch shows you. It's a TXT record — a plain-text\n   note attached to your domain:\n   - **Type:** `TXT`\n   - **Host:** `_bumbleb-verify.example.com` — some registrars want only `_bumbleb-verify` here.\n   - **Value:** the `bb-verify=…` value shown in the hub. Use **Copy** so you get it exactly right.\n\n   Any other TXT records at that host can stay where they are.\n4. Back in the hub, click **Check again**.\n\n## \"Not found yet\"\n\nSeeing this right after you publish the record is normal. DNS changes usually take a few minutes\nand can take several hours to reach everywhere. Wait a while, then click **Check again**.\n\nIf it still fails, the message tells you which of these happened:\n\n- **no record was found** — check the record is published, and that the host is right for your\n  registrar (with or without your domain on the end);\n- **a record was found that doesn't match** — compare the value with the one in the hub; using\n  **Copy** avoids typos;\n- **the lookup timed out** — wait, then **Check again**.\n\n## How to tell it worked\n\nWhen the domain verifies:\n\n- the badge shows **Verified**;\n- the connector becomes **Active** and starts receiving data;\n- your account gets a one-time credit for the domain — see [Credits](/help/hub/credits/).\n\n## Good to know\n\n- **Verify the root domain only.** `example.com` covers `www.example.com` and `shop.example.com`.\n- **Your existing TXT records are safe.** Adding this one doesn't mean removing others at the same\n  host.\n\n## Related\n\n- [Connectors in the hub](/help/hub/connectors/) — where you create the connector this domain\n  belongs to.\n- [Google Tag Manager tag](/help/connect/google-tag-manager/) — finish connecting your website.\n- [Shopify pixel](/help/connect/shopify-pixel/) — finish connecting your store.\n","path":"/help/hub/domains/","section":"Your account","slug":"hub/domains","title":"Verify your domain","url":"https://bumbleb.co/help/hub/domains/"},{"description":"An AI analyst for your data — it answers in plain words and shows how it counted.","gate":null,"markdown":"\nCrunch is an AI analyst. You connect your website, your store or the files you already export, and\nask questions the way you'd ask a colleague — \"why were enquiries weak in March?\" rather than a\nquery. It works out what to count, reads your data, and answers. Every answer says how it read your\nquestion and how it counted, so you can check it rather than take it on trust.\n\n## What it will do\n\n- **Investigate across several datasets.** A dataset is one body of data — your website's visits, an\n  exported orders file, your ad spend. Crunch can bring more than one into the same question.\n- **Plan the hard questions with you**, when a question needs several steps before it can be\n  answered.\n- **Write up what a conversation found**, so you have something to share.\n- **Say plainly what your data can't answer**, instead of guessing.\n\n## What it won't do\n\n- **Invent a figure your data doesn't hold.** If the data can't support a number, Crunch says so.\n- **Give investment advice.**\n- **Change your data.** It only reads.\n\n## Where to start\n\n1. [Sign up and start on the free plan](/help/getting-started/sign-up/) — no card needed.\n2. Give Crunch some data to read, then ask a question in your own words.\n3. Learn to [read an answer](/help/getting-started/reading-an-answer/), so you know what each part\n   tells you.\n\n## Related\n\n- [Sign up and start on the free plan](/help/getting-started/sign-up/) — the first step.\n- [Solve a problem in a conversation](/help/getting-started/solve-a-problem/) — how to ask so each\n  answer sets up the next.\n- [Correcting Crunch](/help/getting-started/correcting-crunch/) — what to do when it reads you wrong.\n","path":"/help/getting-started/what-is-crunch/","section":"Getting started","slug":"getting-started/what-is-crunch","title":"What Crunch is, and what it won't do","url":"https://bumbleb.co/help/getting-started/what-is-crunch/"},{"description":"Only what you send — and exactly what the website tag and Shopify pixel send.","gate":null,"markdown":"\nBefore you connect anything, it's fair to ask what Crunch will actually see. The short answer: only\nwhat you send it. Here is what each way of connecting sends, so you can check it against your own\nprivacy policy and tell your customers with confidence.\n\n## Website tag\n\nThe website tag is a small piece of code on your site that sends events — things visitors do, like\nviewing a page or submitting a form — to Crunch. It sends only the events your site pushes with\n`event: 'bb_event'`, along with the page, the referrer (the page the visitor came from), campaign\nparameters and any fields you add. Nothing else on the page is read.\n\n**Tip:** user IDs can be [hashed in the browser](/help/connect/hash-user-ids/) first, so the raw ID\nnever leaves the visitor's browser.\n\n## Shopify pixel\n\nThe Shopify pixel sends the store's standard events — views, carts and checkouts — with an anonymous\nShopify client ID. It asks Shopify for no customer identity.\n\n## Uploads and linked tables\n\nThe files you upload and the tables you link are read only to answer your questions.\n\n## On-prem sync\n\nOnly the columns your rules let out, in the form the rules set. See\n[Keep sensitive data inside your network](/help/privacy/sensitive-data/) for how those rules work.\n\n## Good to know\n\nEvents are accepted only from domains you have verified. Verifying a domain means proving the website\naddress is really yours, so events claiming to come from anywhere else are not accepted. See\n[Domains](/help/hub/domains/).\n\n## Related\n\n- [Keep sensitive data inside your network](/help/privacy/sensitive-data/) — ways to keep raw personal\n  data on your side.\n- [Hashing user IDs](/help/connect/hash-user-ids/) — hashing IDs in the browser.\n- [Domains](/help/hub/domains/) — verifying the domains your events come from.\n","path":"/help/privacy/what-we-collect/","section":"Privacy \u0026 security","slug":"privacy/what-we-collect","title":"What we collect","url":"https://bumbleb.co/help/privacy/what-we-collect/"},{"description":"Your website, your store, your files, your database, your data lake — and the way to bring each one in.","gate":null,"markdown":"\nBefore you connect anything, it helps to know which way in fits the data you have. What visitors do\non your site arrives as a live stream of events; what's in a spreadsheet arrives as an upload; what\nsits in a database or data lake can be synced or read where it lives. Find your data in the table\nbelow and follow the link.\n\n## Pick your route\n\n| Your data | Route | Plan |\n|---|---|---|\n| What visitors do on your website or web app | [Google Tag Manager tag](/help/connect/google-tag-manager/) | every plan |\n| What shoppers do in your Shopify store | [Shopify pixel](/help/connect/shopify-pixel/) | every plan |\n| Anything you can export — orders, ad spend, reviews, listings | [Upload a file](/help/connect/upload-files/) | every plan |\n| A database behind your firewall | [On-prem sync connector](/help/features/on-prem-sync/) | Team |\n| Tables already in your Postgres or Iceberg data lake | [Your own data lake](/help/features/external-connections/) | Team |\n\n\"Behind your firewall\" means a database on your own servers or private network that the internet\ncan't reach directly. A data lake is a large store of tables your team already keeps, often for\nother analytics tools.\n\n## Ask across all of it\n\nEverything lands in your own workspace, so you can ask one question that spans several sources —\nyour site's sign-ups next to your ad spend, say. See\n[Two datasets, one question](/help/use-cases/two-datasets/).\n\n## Good to know\n\n- **Heads up:** the tag and the pixel need a [verified domain](/help/hub/domains/) before any event\n  is accepted. It's worth starting that step first.\n- **Tip:** no ad-platform connector yet? Your spend comes in as an export — see\n  [Bring in your ad spend](/help/connect/ad-spend/).\n\n## Related\n\n- [Website \u0026 ad ROI](/help/use-cases/website-ad-roi/) — the tag plus a spend export, worked\n  through.\n- [Shopify \u0026 e-commerce](/help/use-cases/shopify/) — the pixel plus your store's exports.\n- [Any dataset](/help/use-cases/any-dataset/) — one uploaded file, start to finish.\n","path":"/help/connect/which-route/","section":"Connect your data","slug":"connect/which-route","title":"Which route for which data","url":"https://bumbleb.co/help/connect/which-route/"},{"description":"See your organization's details, owner, contacts and subscription in one place (Team plans).","gate":"team","markdown":"\nWhen someone asks \"who's the owner of our account?\" or \"how many seats are we paying for?\", the\n**Organization** page has the answer. It gathers your organization's details and its subscription on\none screen, so you can check them without changing anything by accident.\n\n## What you'll need\n\n- A Team plan (see [pricing](/pricing/)).\n- To be the owner, or someone who can manage users — admins can by default. Members don't see the\n  page. See [Roles](/help/hub/roles/).\n\n## What it shows\n\n- **Your organization** — its name, its email domain (the part of your work email after the @),\n  your team size, and when it was created.\n- **People** — the organization's owner and its contacts.\n- **Your subscription** — its status, the number of seats (a seat is one person on the plan), the\n  billing cycle, and any pending change, such as a seat change scheduled for later.\n- **Billing contact** — who receives billing information.\n\n## Good to know\n\nThis page is **read-only**: it shows these details but you can't edit them here. To change your\nplan or seats, use [Subscriptions](/help/hub/subscription/).\n\n## Related\n\n- [Upgrading, downgrading and cancelling](/help/hub/subscription/) — change your plan or seats.\n- [Seats](/help/hub/seats/) — how seats are counted and assigned.\n- [Roles and what each can do](/help/hub/roles/) — what the owner can do that others can't.\n","path":"/help/hub/organization/","section":"Your account","slug":"hub/organization","title":"Your organization","url":"https://bumbleb.co/help/hub/organization/"},{"description":"Your account, your organization and what your role lets you do, at a glance.","gate":null,"markdown":"\n**Profile** is where you check who Crunch thinks you are and what you're allowed to do. It's the\nfirst place to look when something you expect isn't there — a page missing from the hub, say — and\nyou want to know whether it's your role.\n\n## What it shows\n\n**About you:**\n\n- your name and email;\n- your sign-in method — how you sign in to Crunch;\n- whether your email is verified;\n- when you joined.\n\n**Your membership:**\n\n- your organization;\n- your role — owner, admin or member;\n- what your role lets you do.\n\n## Close your account\n\nIf you're the owner, the **Close account** section is at the bottom of Profile. Closing your organization\ncancels your subscription and ends access for everyone in it. Any credits you have left are\nforfeited, and connectors stop collecting at once. If you only want to stop paying, switch to the\nFree plan under [Subscriptions](/help/hub/subscription/) instead, and you keep your account and\nyour data.\n\nTo close:\n\n1. Click **Close Account**.\n2. Type your organization's name to confirm. A confirmation code is emailed to you and to every\n   admin who can manage subscriptions.\n3. Enter the code (it expires in 10 minutes) and confirm.\n\n**If you're an admin** and got that email, open your Profile: while the request is open, a\n**Close account** section shows who asked and a box for the code. Entering it closes the\naccount; ignoring it lets the code expire. Only the owner can start a closure.\n\nAfterwards, the sign-in page points people to support. To reopen a closed account, email\n[support@bumbleb.co](mailto:support@bumbleb.co) from the owner's address.\n\n## Good to know\n\nApart from closing the account, nothing on this page can be changed. If your role or capabilities need to change, ask an owner —\n[Roles](/help/hub/roles/) explains what each role can do and who can change it.\n\n## Related\n\n- [Roles and what each can do](/help/hub/roles/) — understand what your role allows.\n- [Tour of the hub](/help/hub/tour/) — what each page in the hub is for.\n","path":"/help/hub/profile/","section":"Your account","slug":"hub/profile","title":"Your profile","url":"https://bumbleb.co/help/hub/profile/"},{"description":"See what you've used this month against your own plan, and what's using the most.","gate":null,"markdown":"\nEvery question you ask Crunch uses some of your plan's monthly included usage. **Usage** shows how\nmuch you've used this month and where it went, so you can tell whether you're on track or likely to\nneed [credits](/help/hub/credits/) before the month resets.\n\n## Your usage is yours\n\nUsage is measured per person, against your own plan. A colleague's use never counts against yours,\nso a busy teammate can't use up your month.\n\n## What it shows\n\n- **Plan allowance** — how much of this month's included usage you've used, and when it resets.\n- **Credits** — how many you've spent and how many are left. Credits pay for usage beyond what your\n  plan includes.\n- **By agent** — turns and credits for each kind of work: analysis, planning, reports, and the data\n  console. A turn is one question and its answer.\n- **By model** — turns and credits for each model, the AI behind Crunch's answers.\n\n## Good to know\n\nBigger models and higher effort use more per question. If your allowance is going faster than you\nexpected, **By model** shows whether a bigger model is the reason. See\n[Choosing a model](/help/features/models/) for how to pick.\n\n## Related\n\n- [Credits](/help/hub/credits/) — what happens when your included usage runs out.\n- [Team usage](/help/hub/team-usage/) — everyone's usage by name (Team plans).\n- [Choosing a model](/help/features/models/) — balance answer quality against usage.\n","path":"/help/hub/usage/","section":"Your account","slug":"hub/usage","title":"Your usage","url":"https://bumbleb.co/help/hub/usage/"}]}