Any MCP-compatible appRead live data
- The tables and views you connect, with their fields
- Your own metrics and dimensions, defined once
- CRM, sales or offline data by date
- Next to your marketing sources in one question
Connect once, use it two ways
Connect BigQuery to Jepto once. What you build in Jepto and what you ask in Claude or ChatGPT read the same data, so the numbers match wherever you look.
Add the Jepto MCP server and ask about BigQuery in plain English, across all your brands. Included on every plan.
and any MCP-compatible appJepto runs a hosted MCP server at mcp.jepto.com that puts your BigQuery tables in Claude, ChatGPT, Cursor or any MCP-compatible app, next to your ad platforms and analytics. Your tables become named metrics and dimensions, so the AI asks for revenue or deals by month and nobody writes SQL. It reads your BigQuery data and never changes it.
Paste the URL into your AI assistant, or find Jepto in Claude's connector directory or ChatGPT's Plugins page. Sign in to Jepto once and every conversation can use your live BigQuery data.
Works with
OAuth sign-in. Each person sees only the brands their Jepto login can.
Nothing to set up from chat for BigQuery.
Nothing. The Jepto MCP reads your BigQuery data and never changes it.
Paste one into Claude or ChatGPT once Jepto is connected. Swap [brand] for one of yours.
“What fields are in [brand]'s CRM table, and which ones can be summed?”
“Show closed deals from the CRM table by month next to Google Ads spend for this year.”
“Which lead sources in our offline sales data produced the most revenue last quarter?”
“Work out cost per sale for last month using total ad spend and the sales in our BigQuery table.”
Google's own server is the right tool for a data team writing SQL. Jepto suits marketers: your tables become named metrics and dimensions that sit next to the ad platforms, so nobody writes a join.
| Compared on | Jepto MCP | Google's BigQuery MCP server | Community servers |
|---|---|---|---|
| Setup | Hosted. Paste one URL, or pick Jepto in the Claude or ChatGPT directory, then sign in. | Enable it in your Google Cloud project and connect with Google credentials. | You install, host and update it, and supply your own API credentials. |
| Accounts | Every BigQuery account you've connected, across all your brands, in one connection. | Projects your Google login can reach. | Usually one login or token at a time. |
| Other channels | Ask across BigQuery and every other connected source in the same question. | Anything already in BigQuery. | One platform per server. You combine servers yourself. |
| Brand context | Reads each brand's details and knowledgebase, so answers reflect what the business sells and what it's aiming for. | None. | None. |
| Field guidance | Ships a query guide the AI reads first, so it picks the right BigQuery fields and says what isn't available. | The AI writes SQL against your raw tables. | Raw API fields. The AI works out what they mean. |
| Changes | Read-only. It queries the tables you connect and never writes to BigQuery. | Runs the SQL it's given. | Varies. Some write to your account with no confirm step. |
| Cost | Included on every Jepto plan. BigQuery's own query costs still apply to your project. | Free to connect. Query costs apply. | Free code. Your time to run and maintain it. |
There's no fixed BigQuery field list. You connect a table or view, and its columns become metrics and dimensions that work the same way in Reports, Data Chat and the Jepto MCP.
| How you connect | Sign in with Google Cloud, then choose a project, dataset and table or view. You need BigQuery access to the project. |
| Fields | The columns in your table. You choose each one's type and aggregation, and hide the ones you don't need |
| Use it in | Reports, Data Chat and the Jepto MCP |
| AI assistants | Claude, ChatGPT, Cursor and any MCP client, at https://mcp.jepto.com |
| Changes from chat | None. The MCP reads your data and never changes it |
| Numbers (INT64, FLOAT64, NUMERIC, BIGNUMERIC) | Metrics. Choose Sum, Average, Minimum or Maximum for each one. |
| Text (STRING, BYTES) | Dimensions you can group and filter by. |
| True or false (BOOL) | Dimensions you can group and filter by. |
| Dates and times (DATE, DATETIME, TIMESTAMP, TIME) | Date fields. Pick the one that report date ranges should use. |
| Other types | Read as text dimensions. |
| Tables per data source | One table or view. To combine tables, create a view in BigQuery and connect the view. |
| SQL | None needed. Jepto builds each query from the columns you've mapped. |
| Copies of your data | None. Jepto queries BigQuery when a report, Data Chat or MCP request runs. |
| Partitioned tables | Supported. Report date ranges filter on the partition column, and very wide ranges are rejected. |
| One schema for many brands | Make the data source a template, then point each brand at its own project, dataset and table in Client Config. |
Reports and Data Chat all use the same columns you map.
Describe the report you want and Jepto builds it from your BigQuery data, then writes an analysis of what changed and why.
Explore reportingAsk questions about your BigQuery data in plain English and get answers, tables and charts back in seconds.
Explore Data ChatConnect Google Cloud once, then register a table or view for each brand.
In Data Sources, open Connections, add Google Cloud Platform and sign in with a Google account that can query the project.
Add a BigQuery data source, pick the project, dataset and table or view, then click Load fields.
Confirm each column's type, choose an aggregation for number columns, hide what you don't need and pick the date field. Then save.
Your data is ready in Reports and Data Chat. Connect https://mcp.jepto.com to your AI assistant to ask about it from there too.
About BigQuery in Jepto and the Jepto MCP.
Yes. Jepto hosts an MCP server at https://mcp.jepto.com that gives Claude, ChatGPT, Cursor and other MCP-compatible apps access to the BigQuery tables you connect to Jepto. There's nothing to install: you sign in to Jepto with OAuth, and the assistant can query the columns you've mapped.
Yes. Google Cloud hosts a BigQuery MCP server that lets an AI write and run SQL in your projects. Jepto's server works at a different level: it exposes your connected tables as defined metrics and dimensions next to your marketing data, so marketers can ask questions without SQL.
No. Through the Jepto MCP, BigQuery data is read-only: the assistant can query your table but can't change anything in BigQuery.
No. Jepto queries BigQuery live when a report, Data Chat or MCP request runs. Nothing is copied into Jepto, so your reports are as current as the table.
No. You map columns and choose aggregations, and Jepto builds the query. If you need joins or calculated columns, create a view in BigQuery and connect the view.
Yes. Turn on Make this data source a template when you create it. The columns, types and aggregations are shared, and each brand points at its own project, dataset and table in Client Config, so one master report works for all of them.
The Google account you connect needs to see the project and read the dataset, for example with BigQuery Data Viewer on the dataset. If a project or table doesn't appear in Jepto, that account doesn't have access to it.
No. Data Warehouse pipelines write your marketing data into BigQuery. A BigQuery data source reads a table you own into Jepto so you can report on it.
No. The Jepto MCP is included on every plan with no usage limits. Check that your AI assistant's plan lets you add connectors.
Related Articles
Start asking your data smarter questions today. Connect your first data source in minutes.
