All insights
Analytics
3 min read29 July 2026Nathan Mzumara

GA4 Just Shipped Source Groups. AI Traffic Is Finally Legible.

GA4 Just Shipped Source Groups. AI Traffic Is Finally Legible.

Google Analytics has quietly added two features that most teams have wanted for years: a Source Group dimension and hostname filters. Together they roll messy source/medium values into clean platform categories and let you exclude spam domains, so your reports finally read the way your channels actually behave.

If you have ever spent a Friday afternoon merging chatgpt.com, chat.openai.com and openai.com into one row in a spreadsheet, this is the fix. In my opinion it is one of the more useful reporting-layer changes GA4 has shipped, precisely because it does the boring work for you.

What actually changed

GA4 now offers Source Group as a native dimension. Instead of a long tail of raw referrers, traffic gets bucketed into logical platform groups, which makes emerging AI referral traffic from ChatGPT, Perplexity and Gemini countable in a single view.

Hostname filters do the other half of the job. They let you exclude self-referrals, staging environments and spam domains at the report level, so your numbers reflect real visits rather than noise.

When it landed and what it covers

The rollout has been gradual across GA4 properties rather than a single flag-day launch. Crucially, from my observation the Source Group dimension applies retroactively, so historic data reorganises into the new groupings without you re-tagging anything.

That retroactive benefit matters. You get clean trend lines back through the period you already care about, not just a fresh start from the day you noticed the feature.

How it works

The mechanism is straightforward. GA4 maps each raw source and medium to a broader Source Group, then hostname filters remove entries you do not want counted before the report renders.

Raw source/medium (before)Source Group (after)
chatgpt.com / referral
chat.openai.com / referral
ChatGPT
google / organic
google / cpc
Google
perplexity.ai / referralPerplexity

Illustrative example: how scattered raw values collapse into a single Source Group.

Set it up in four steps

  1. Open a report or exploration and add Source Group as your primary dimension in place of source/medium.
  2. Apply a hostname filter to include only your production domains, which removes staging and self-referral noise.
  3. Build an exclusion list for known spam or bot domains so they never enter the report.
  4. Save the view as a shared report so the whole team reads the same clean numbers.

What it means for growth teams

The practical win is legibility. Cross-channel reporting stops being a manual reconciliation exercise, and AI-assisted discovery becomes a line you can trend, defend and report to the board.

I think this quietly raises the bar for how teams measure emerging channels. If you have been arguing that AI referrals are too small to track, the data was probably just too fragmented to see. For the wider context on why direct measurement matters as platforms shift, see our note on how regulation is reopening direct discovery channels.

Verify the current behaviour against the primary docs before you rely on it: the Google Analytics developer documentation and the official GA4 Help Centre are the sources of truth as the rollout continues.

Concrete action: rebuild your channel report around Source Group today, add hostname filters, and check whether your AI referral traffic has been hiding in the long tail all along.

Tags

GA4Google AnalyticsAI referral trafficattributionreportingsource groupshostname filterscross-channel measurement

The Discovery Digest · Every Friday

Stay ahead of AI Search

Ten updates a week across ChatGPT, Claude, Gemini, Perplexity, Copilot, Grok and Google AI Overviews, with the questions worth asking.

Free10 updates weeklyUnsubscribe anytime