For the past couple of years, visits from AI assistants have been hard to spot in your analytics. Some showed up as ordinary referrals, some landed in “Direct,” and many disappeared into “Unassigned.” Two recent changes from Google make it much easier to measure AI search traffic on a WordPress blog: Gemini has started adding UTM tags to the links it sends out, and Search Console now reports how often your pages appear in AI Overviews and AI Mode.
Neither tool tells the whole story on its own. This guide shows you how to set up Google Analytics 4 (GA4) to catch AI referrals, how to read Search Console’s AI reports without misreading them, and how to fold both into a 30-minute monthly check.
What Changed: Gemini UTM Tags and Search Console AI Reports
WordPress publishers can now measure AI search traffic more reliably because Gemini adds tracking parameters to some outgoing links, and Search Console shows impressions from Google’s AI search features.
At the start of October, site owners noticed that Gemini had quietly started adding UTM parameters to outgoing links. Google hasn’t documented the change yet, so the exact tag values and when they’re applied aren’t confirmed. In practice, that means you should match on anything containing “gemini” rather than on one exact value.
Separately, Google finished rolling out AI performance reports in Search Console to sites worldwide at the end of August. They show impressions from AI Overviews, AI Mode and Discover’s generative AI features, broken down by page, country and date. In October, Google also added a new multimodal search filter to the performance reports.
There are two important caveats:
- The Search Console AI reports show impressions, not clicks. They tell you that your page appeared in an AI answer, not whether anyone visited. For clicks, you still need GA4.
- Position data is limited. Google has acknowledged that its AI reporting reflects where the answer block sits, not where your link sits inside it, and that the reporting is incomplete.
These changes sit alongside Google’s other recent moves to send more people from AI answers to websites, such as Preferred Sources in AI Mode, which we covered last week.
Setting Up GA4 to Measure AI Search Traffic
To measure AI search traffic in GA4, create a custom channel group with an “AI Search” channel that matches the source names of Gemini, ChatGPT, Perplexity, Copilot and Claude, and place it above the default Referral channel.
By default, GA4 spreads AI visits across several buckets. A visit with a referrer usually lands in “Referral.” A tagged link with a source but no recognized medium can end up in “Unassigned.” A click from an app that strips the referrer often counts as “Direct.” A custom channel group pulls the visits GA4 can identify into one place.
Here’s what to look for from each assistant:
AI assistant referral sources at a glance
| Assistant | Source or referrer to match | What to watch for |
|---|---|---|
| Google Gemini | gemini.google.com |
Some links now carry UTM tags; exact values aren’t documented yet |
| ChatGPT | chatgpt.comchat.openai.com |
Tags links in web-grounded answers, though some links have recently been reported missing their tags |
| Perplexity | perplexity.ai |
Usually appears as a referral |
| Microsoft Copilot | copilot.microsoft.com |
Clicks from some Windows and app surfaces may arrive without a referrer |
| Claude | claude.ai |
Usually appears as a referral |
Referrer behavior can change without notice, so confirm these sources against your own GA4 data.
Create an AI Search Channel Group in GA4
A custom channel group is the best way to do this because it applies to your past data as well as new visits, so you can see trends straight away.
- Open channel groups. In GA4, go to Admin, then Data display and Channel groups.
- Copy the default group. You can’t edit the default channel group, so make a copy and give it a name like “Channels with AI Search.”
- Add an AI Search channel. Click Add new channel, name it “AI Search,” and set the condition to Source matches regex.
- Paste this pattern:
gemini|bard\.google|chatgpt|openai|perplexity|copilot|claude\.ai - Move it above Referral. Save the channel, then reorder the list so AI Search sits above Referral and Organic Search. GA4 assigns each visit to the first channel it matches, so order matters.
- Check your results. In Reports, open Acquisition and then Traffic acquisition, and switch the primary dimension to your new channel group.
Build an Exploration Segment for Deeper Analysis
Channel groups can’t match on page referrer, but explorations can. For a closer look at how AI visitors behave, go to Explore, start a blank exploration and create a session segment where Session source or Page referrer matches the same pattern. Compare that segment with your organic search visitors on engagement rate, pages per session and key events like newsletter signups.
Reading the Search Console AI Reports
Search Console’s AI reports show how often your pages appear in AI Overviews, AI Mode and Discover’s AI features, which helps you find pages that AI search is surfacing but that aren’t getting many visits.
When you open the reports, focus on three things:
- Which pages get the most AI impressions. These are the posts Google’s AI features lean on most. Make sure they’re accurate and up to date.
- How impressions change over time. A sudden drop on a key page can mean a newer or clearer source has replaced you.
- Which countries and devices they come from. This tells you whether your AI visibility matches the audience you actually serve.
Search Console also has a “Search generative AI” setting that lets you opt your site out of AI Overviews and AI Mode. Most blogs shouldn’t use it, because opting out removes those impressions and any visits they send. Know it exists, but think carefully before you switch it on.
Your Monthly AI Traffic Check
A good monthly AI traffic check pairs Search Console’s impression data with GA4’s visit and conversion data, so you can see both where your content appears and what those appearances are worth.
Neither tool is enough on its own. Search Console covers only Google, while GA4 covers clicks from every assistant but not the answers people read without clicking. Set aside about 30 minutes at the start of each month for these four steps:
- Record your GA4 AI totals. In Traffic acquisition, note sessions, engaged sessions and key events for the AI Search channel, broken down by source.
- Compare top pages. Export your top 25 landing pages from the AI Search channel and your top 25 pages from Search Console’s AI report. Pages with lots of AI impressions but few visits are your best candidates for clearer headings, stronger answers or a more compelling reason to click.
- Watch branded searches. In the standard Search Console performance report, filter for queries containing your brand name. When AI impressions rise, branded searches often rise too, as readers look you up after seeing your name in an answer.
- Spot-check pages that dropped. For any important post that lost AI visits, ask Gemini, ChatGPT and Perplexity the question it answers and see who gets cited instead. Our step-by-step AI citation audit walks through this in detail, and our comparison of which AI platforms cite WordPress sites most can help you decide where to start.
Keep in mind that not every AI mention shows up as a visit. Sometimes an assistant links to your post without naming you, and sometimes it names you without a link. We explain the first case in our post on ghost citations.
Adding Bing Webmaster Tools to the Picture
Bing Webmaster Tools fills a gap Google’s tools leave open: its AI Performance report shows where your content is cited in Microsoft Copilot and Bing’s AI answers, and groups the questions behind those citations into topics.
Microsoft added intents, topics, citation share and period comparisons to the AI Performance report earlier this year. Topics are especially useful because they show the broader questions people ask, not just isolated keywords. Note that citation share measures how often you’re cited for a query, not how much traffic you get.
To fold Bing into your monthly check:
- Verify your site. If you haven’t already, sign in to Bing Webmaster Tools and import your site from Google Search Console. It’s the quickest way to get set up.
- Review your top topics. In the AI Performance report, note the topics where you’re cited most and any where your citation share is falling.
- Match topics to GA4 visits. GA4 shows which pages Copilot visitors land on, but not what they asked. Matching your top Copilot landing pages against Bing’s topics helps you work out which questions actually bring people to your site.
For more on improving how Copilot cites you, see our guide to Bing AI visibility.
Frequently Asked Questions
Does Search Console show clicks from AI Mode?
Not yet. The AI performance reports show impressions from AI Overviews, AI Mode and Discover’s AI features, but not clicks. Use GA4 to count the visits that actually reach your site.
What UTM values does Gemini use?
Google hasn’t documented them. Match on any source containing “gemini” so your reports keep working if the values change.
Why does some AI traffic still show up as Direct?
Some apps and in-app browsers don’t pass a referrer, and not every link carries UTM tags. Those visits can’t be identified as AI traffic, so treat your AI Search channel as a minimum count rather than an exact total.
Will a custom channel group change my historical data?
It changes how past data is grouped in reports, but it doesn’t alter the underlying data, and your default channel group stays as it was. You can compare the two at any time.
Should I opt out of AI Overviews and AI Mode?
For most blogs, no. Opting out removes your impressions and any visits from those features. It’s mainly worth considering if AI answers are fully replacing visits you rely on, such as for paid content.