Can one AI agent read both Google Ads and GA4 in the same analysis?

Yes — cross-surface questions are the main reason to use one. It reads Google Ads, Meta, TikTok, DV360, CM360, GA4, Search Console, BigQuery and Shopify, and can reason across them in a single run.

How does it calculate ROAS differently from the ad platforms?

It writes SQL against your own BigQuery warehouse to join ad cost with your actual revenue data, rather than relying on each platform's click-time-attributed conversion value, which doesn't account for refunds or returns.

Can analysts use it without being able to change campaigns?

Yes — the Member role can create tasks and view results but cannot authorize integrations or approve any write. Approval rights sit with Editors and Admins.

Can it produce our recurring stakeholder report automatically?

Yes — save the goal that produces it and schedule it on a cron. It re-runs against live data each time rather than reproducing a stale snapshot, and can output a written report or a slide deck.

Does it replace our BI tool?

No — it queries your warehouse to answer specific questions and produce reports, but it isn't a dashboarding layer. Teams generally keep BI for standing dashboards and use the agent for the questions that need reasoning across systems.

For

AI ad ops for in-house marketing teams — one agent across ads, analytics and your warehouse

TL;DR

An in-house team's hardest questions are the cross-surface ones — is paid cannibalising organic, does this campaign actually make money after refunds, which channel deserves next quarter's budget. Those need one system that reads the ad platforms, the analytics and your own warehouse together, which is exactly the shape Agent Planners is built around.

The questions that span systems

Question	What it has to read at once
"Are we bidding on keywords we already rank #1 for organically?"	Google Ads keyword spend + Search Console query positions
"What's our real ROAS after refunds and returns?"	Ad cost across platforms + your own BigQuery or store revenue
"Which channel should get next quarter's incremental budget?"	Google Ads, Meta, TikTok performance + GA4 conversion paths
"Did last week's landing page change help or hurt paid conversion?"	GA4 behaviour + Google Ads campaign performance

Each of these is unanswerable inside any single platform's own UI — which is why they usually end up in a spreadsheet someone rebuilds every month.

True ROAS from your own numbers

Every ad platform reports conversion value using its own click-time attribution, before refunds, returns or the discount that shaved 20% off the order. The agent writes its own SQL against your BigQuery warehouse to join ad cost with what actually landed — which is the difference between a dashboard number and a number you'd defend in a budget meeting. See true ROAS from your own data.

Reporting that doesn't eat a day a month

The recurring stakeholder report is the clearest win: schedule the goal that produces it and the agent re-runs it against live data on a cron rather than someone rebuilding the same deck each cycle. Output can be a written report or a generated slide deck — reports & decks covers the formats, and scheduling covers the cadence.

Access that matches team structure

Analysts can run tasks and read results on the Member role without ever holding the ability to authorize an integration or approve a write to a live campaign. Editors approve changes; Admins control members and billing. That separation lets you widen access to the analysis without widening access to the account.

Platforms it reads natively

Ads — Google Ads, Meta, TikTok, DV360, Campaign Manager 360.
Analytics — GA4, Search Console, BigQuery.
Commerce — Shopify.
Publisher side — Google Ad Manager, AdSense, AdMob.
Everything else — 250+ SaaS apps via Composio, plus your own tools over MCP.

Frequently asked questions

Can one AI agent read both Google Ads and GA4 in the same analysis?: Yes — cross-surface questions are the main reason to use one. It reads Google Ads, Meta, TikTok, DV360, CM360, GA4, Search Console, BigQuery and Shopify, and can reason across them in a single run.
How does it calculate ROAS differently from the ad platforms?: It writes SQL against your own BigQuery warehouse to join ad cost with your actual revenue data, rather than relying on each platform's click-time-attributed conversion value, which doesn't account for refunds or returns.
Can analysts use it without being able to change campaigns?: Yes — the Member role can create tasks and view results but cannot authorize integrations or approve any write. Approval rights sit with Editors and Admins.
Can it produce our recurring stakeholder report automatically?: Yes — save the goal that produces it and schedule it on a cron. It re-runs against live data each time rather than reproducing a stale snapshot, and can output a written report or a slide deck.
Does it replace our BI tool?: No — it queries your warehouse to answer specific questions and produce reports, but it isn't a dashboarding layer. Teams generally keep BI for standing dashboards and use the agent for the questions that need reasoning across systems.