How to get your Google Analytics 4 data ready for agentic AI

Sync GA4 reports and custom reports with Airbyte so AI agents can analyze traffic, conversions, and engagement from a governed warehouse copy.

|
Nov 16, 2026

Summarize with AI:

Airbyte syncs Google Analytics 4 report streams and custom reports from the Data API into your warehouse, and the Airbyte context layer lets agents read core GA4 reports directly through the Airbyte MCP.

Web analytics is a natural fit for AI agents. Questions like "why did signups drop on Tuesday" or "which landing pages bring the most engaged visitors" require slicing the same metrics by many dimensions, and that is mechanical work an agent can do well. GA4 is also a tricky source to query on demand. Reports are aggregated, sampling can apply to large requests, numbers shift as attribution settles, and API quotas cap how much you can ask for.

A warehouse copy of GA4 reports, synced with settings that limit sampling, gives agents stable numbers they can join with ad spend, CRM, and revenue. This guide covers how Airbyte builds that copy.

Why GA4 data matters for agentic AI

GA4 connects marketing activity to on-site behavior:

  • Acquisition. User and traffic acquisition reports show which sources, mediums, and campaigns bring visitors.
  • Engagement. Events, pages and screens, and landing pages show what visitors do.
  • Conversions. Conversion and ecommerce reports show what visitors complete.
  • Audience and technology. Demographic, location, device, browser, and operating system reports explain differences between segments.

In a warehouse, GA4 data becomes the bridge between ad platform metrics and revenue systems.

What agents can do with GA4 data

  1. Explain traffic changes. Break a drop or spike down by source, medium, landing page, device, and country to find where it came from.
  2. Rank landing pages. Compare pages by sessions, engagement, and conversions over a period.
  3. Connect spend to outcomes. Join GA4 sessions and conversions by campaign with Google Ads, Facebook, or Bing spend.
  4. Monitor key events. Track daily counts of important events and flag unusual days.
  5. Answer quick questions. With agent access on, read daily active users or top traffic sources directly without waiting for a sync.

How Airbyte gets GA4 data ready

Authentication

On Airbyte Cloud, sign in with Google OAuth. On self-managed Airbyte, use a service account with the Viewer role on the GA4 property, and enable the Google Analytics API that the connector docs list in the service account's Google Cloud project.

The connector only supports GA4 properties. Universal Analytics properties, identified by a Tracking ID such as UA-123-1, cannot be reported on through the Data API.

Sync modes

The connector supports Full Refresh Overwrite, Full Refresh Append, Incremental Append, and Incremental Append + Deduped.

Streams

The connector ships a large set of preconfigured report streams built on the Data API's runReport method. They cover user and traffic acquisition, events, conversions, pages and screens, landing pages, ecommerce, demographics, and technology, along with daily, weekly, and four-weekly active users and a website overview.

A separate property_metadata stream uses the Admin API to emit one record per property, with a property_id for joins. It requires the Google Analytics Admin API to be enabled in the credentials' Google Cloud project.

Custom reports

You can define custom reports with a name, dimensions, and metrics, plus an optional cohort spec and pivots. Custom reports with pivots use the Data API's pivot method. The start date and Data Request Interval settings do not apply to custom cohort reports.

Accuracy settings

  • Data Request Interval (Days). Ranges from 1 to 364. Smaller intervals reduce the chance that GA4 samples results, and Airbyte recommends 1 day unless you need faster syncs. If your reports use month or yearMonth dimensions, the docs note that very short intervals can produce duplicate entries, and suggest 364 for those cases.
  • Lookback window (Days). Refreshes a number of past days on each run, because attribution changes after the event date and GA4 has processing latency.
  • Keep Empty Rows. Optionally returns rows where all metrics are zero.

Multiple properties

By default, each report produces one stream per property. One Stream per Report consolidates each report into one stream across all configured properties, with property_id in the primary key. It changes stream names, so enabling it on an existing connection requires a full re-sync. Properties added later join the consolidated stream from the current cursor and are not backfilled unless you clear the stream.

Limits

The connector is subject to Google Analytics Data API quotas. Smaller request intervals mean more requests, so balance accuracy and quota use.

How agents reach GA4 data

GA4 is an agent connector in Airbyte Cloud:

  • Agent access (direct reads). Agents can list core reports: website overview, daily, weekly, and four-weekly active users, traffic sources, pages, devices, and locations. Reads use the credentials stored in Airbyte and are read-only. They count toward Data API quotas.
  • Search. When a connection syncs GA4 to an agent-connector destination, selected streams can be indexed for search.
  • Warehouse SQL. Synced to BigQuery or Snowflake, the full set of report streams and custom reports is available to agents through read-only SQL, with history.

Agents connect through the Airbyte MCP at https://mcp.airbyte.com/mcp. Direct reads and searches are free up to 100,000 combined per organization per month, and the context layer is available only in Airbyte Cloud.

What this unlocks

  • Traffic and conversion analysis from a stable, typed copy.
  • Fewer sampled numbers through short request intervals.
  • Consistent recent data through the lookback window.
  • Consolidated multi-property reporting.
  • GA4 joined with ad spend and revenue for end-to-end questions.

Getting started

  1. Create a GA4 source in Airbyte Cloud, sign in with Google, and enter your property IDs.
  2. Keep the Data Request Interval low and set a lookback window.
  3. Add custom reports for the dimensions your team uses.
  4. Sync to BigQuery or Snowflake, then connect your agent to the Airbyte MCP.

Try Airbyte free or talk to sales about analytics data for agents.

Frequently asked questions

Does the Airbyte GA4 connector work with Universal Analytics properties?

No. The connector uses the Google Analytics Data API, which only reports on GA4 properties. A property with a Tracking ID such as UA-123-1 is a Universal Analytics property and cannot be synced with this connector.

How do I reduce sampling in GA4 data synced by Airbyte?

Lower the Data Request Interval. Airbyte recommends 1 day, because smaller requests are less likely to exceed Google's thresholds that trigger sampling. Larger intervals, up to 364 days, sync faster but risk sampled results.

Why do recent GA4 numbers change after a sync?

Attribution can change after the event date and GA4 has processing delays. The connector's lookback window refreshes a set number of past days on every run so recent numbers stay consistent.

Can I sync several GA4 properties into one table?

Yes. Enable One Stream per Report to create one consolidated stream per report covering all configured property IDs, with property_id in the primary key. Turning it on for an existing connection changes stream names and requires a full re-sync.

Can agents query GA4 directly through Airbyte?

Yes, for core reports. With agent access on in Airbyte Cloud, agents can list reports such as website overview, active users, traffic sources, pages, devices, and locations. For custom reports and history, sync to a warehouse.

Integrate with 700+ apps using Airbyte

Move data from 700+ sources into warehouses, lakes, and beyond. Set up pipelines in minutes with pre-built connectors and the Connector Builder.