M
Ads / Analytics

Metabase AI Integration

Connect Metabase to BeforeQuery for grounded AI answers and agent actions across Ads / Analytics.

Connect Metabase
What is it?

About Metabase

Metabase is one of the Ads / Analytics platforms BeforeQuery connects to. Once linked, the platform's data becomes a first-class source for grounded AI answers and a target for policy-gated agent actions — the same identity, approval, and audit-log guarantees every other integration ships with.

Setup

How to connect Metabase

  1. Step 01

    In the BeforeQuery dashboard, go to Integrations → Ads / Analytics and select Metabase.

  2. Step 02

    Sign in to Metabase with an account that has permissions to grant BeforeQuery the Ads / Analytics scopes required.

  3. Step 03

    Confirm the connection. Metabase is now available as a source and action target for every playbook in your workspace.

Explore

Related integrations

A
Amplitude

Amplitude is one of BeforeQuery's Ads / Analytics integrations. Connect it once and every agent in your workspace can read, reason over, and act on Amplitude data with your policies + audit trail.

BW
Bing Webmasters

Bing Webmasters is one of BeforeQuery's Ads / Analytics integrations. Connect it once and every agent in your workspace can read, reason over, and act on Bing Webmasters data with your policies + audit trail.

B
Bitly

Bitly is one of BeforeQuery's Ads / Analytics integrations. Connect it once and every agent in your workspace can read, reason over, and act on Bitly data with your policies + audit trail.

How BeforeQuery uses Metabase

What the agent does with your Metabase connection

  • 01

    Read campaign performance, spend, and attribution data from Metabase into the Context Graph.

  • 02

    Answer performance questions grounded in Metabase data ("what's my CAC?", "which channel converts best?").

  • 03

    Trigger playbooks on Metabase anomalies (spend spike, CTR drop, conversion drop) — page the right owner with context.

  • 04

    Draft campaign reports, executive dashboards, and week-over-week updates grounded in Metabase data.

Use cases

What teams actually do with the Metabase integration

Weekly performance digest

Every Monday, BeforeQuery pulls the week's Metabase data, compares to prior week + prior year, flags anomalies, and drafts a digest for the marketing leader. Delivered before the standup; no manual report-building.

Anomaly-driven alerts

Metabase spend spikes 40% over baseline → BeforeQuery investigates (which campaign, which region, what changed), routes to the owning team with context, and pauses the offender if configured to.

Ad-hoc performance Q&A

CMO asks "what's our CAC on Google Ads this month?" → BeforeQuery pulls Metabase data, cross-references with new customers in CRM, computes CAC. Answer in seconds; drill-down link to source data.

Answer questions grounded in Metabase

Employees and customers ask Metabase-related questions in Slack, Teams, or the widget — BeforeQuery answers with the exact Metabase record shown alongside the response. No context switching, no "log into Ads / Analytics to check" round-trips.

Metabase — Frequently Asked Questions

Common questions about connecting BeforeQuery to Metabase.

Yes, gated by approval matrix. Read-only Q&A is autopilot. Pause actions typically require the marketing owner's approval. Budget changes always require an approver with the current spend + proposed spend shown.
Standard OAuth 2.0 where Metabase supports it, otherwise API-key or bearer-token auth. Credentials are stored encrypted at rest (AES-256-GCM) per workspace, never shared across customers, and rotated on request. The connection is scoped to the least-privilege set of scopes each playbook needs — you approve the scope list on install.
No. Your Metabase data feeds only your workspace's own agents and answers. Nothing is used to train a shared model, and nothing crosses workspace boundaries. Bring-your-own-model is available on Enterprise if you want to pin inference to your own OpenAI / Anthropic / self-hosted deployment.
Yes. Every Metabase integration is scoped at install time — you pick which resources, users, or record types are visible. You can further restrict per-playbook: a single playbook only touches the specific Metabase objects it needs. Scope changes take effect on the next sync.
Every write is logged with the acting user, the playbook that fired it, the exact operation, the target entity IDs, and the timestamp — all in the immutable audit log. Writes are rehearsable in simulation mode against a copy of live data before they touch Metabase for real.

Ready to connect Metabase?

Start free. No credit card required. Connect your first integration in under 5 minutes.

Get started free