> ## Documentation Index
> Fetch the complete documentation index at: https://support.detrics.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Connect to Claude

> Connect your Detrics BigQuery warehouse to Claude and ask questions about all your marketing data in plain language

<Info>**Last updated:** May 31, 2026</Info>

Once your marketing data lives in BigQuery, you can connect it to **Claude** and ask questions about it in plain language. No SQL, no exports. Claude reads the real rows in your warehouse and answers across every platform, account, and country at once.

It is easy to set up, and you only do it once.

## How to connect

<Steps>
  <Step title="Create an OAuth client in Google Cloud">
    In your GCP project (with the BigQuery API enabled), go to **Google Auth Platform → Clients → Create client** and choose **Web application**.

    * Authorized JavaScript origins: `https://claude.ai`
    * Authorized redirect URI: `https://claude.ai/api/mcp/auth_callback`

    Copy the **Client ID** and **Client Secret**.
  </Step>

  <Step title="Add the BigQuery connector in Claude">
    Open Claude's connector directory, select **Google Cloud BigQuery**, and paste the Client ID and Client Secret.
  </Step>

  <Step title="Authorize and you're done">
    Sign in with a Google account that has access to your BigQuery project and approve. Claude is now connected to your warehouse and ready to answer.
  </Step>
</Steps>

That's the whole setup. From here, just ask.

## What you can do

* Ask questions in plain language across all your platforms, accounts, and countries.
* Let Claude write and run the SQL for you, then explain the result.
* Walk through multi-step analysis, where each answer builds on the last.
* Draft dashboards, reports, and summaries straight from live data.
* Investigate trends, anomalies, and "why did this change" questions in one conversation.

## Why this matters

### The problem with scattered data

When your numbers live across dozens of spreadsheets and platform dashboards, a few things break:

1. **No single source of truth.** The same metric shows different values depending on which file you open.
2. **Misaligned teams.** Marketing, BI, and finance each work from their own copy and reach different conclusions.
3. **Incomplete reports.** Pulling everything together by hand is slow, so reports end up partial and you lose the big picture.
4. **Slow decisions.** By the time the data is gathered and cleaned, the moment to act has passed.
5. **Unreliable AI answers.** An AI with no grounded data guesses and hallucinates, which leads to bad calls dressed up as confident ones.

### The solution: one centralized warehouse

Detrics syncs every platform and account into a single BigQuery warehouse. The data arrives normalized, deduplicated, and refreshed on the schedule you set. That gives every team and every tool one consistent, current source to work from.

### Why connecting it to Claude is powerful

A centralized warehouse is the foundation. Claude on top of it is the leverage:

* **Grounded answers.** Claude reads your actual rows, so its reasoning is anchored to real numbers instead of guesses.
* **Chain of reasoning.** It can break a hard question into steps, query each one, and combine the results into an answer you can trust.
* **Dashboards on demand.** Describe the view you want and Claude builds the query and the breakdown for you.
* **Real-time decisions.** Ask a question and get the answer now, against data that is already up to date.

### Why Detrics

The hard part is not Claude. It is getting clean, unified, current data into one place. Detrics handles the platform authentication, the API maintenance, the column normalization, the deduplication, and the scheduled refresh. Claude always queries data that is complete and trustworthy, so the answers are too.

<Card title="Set up your warehouse first" icon="database" href="/bigquery/quickstart">
  New to the Data Warehouse? Start here to get your data into BigQuery, then come back to connect Claude.
</Card>