How to Track ChatGPT and Gemini Bookings in Google Analytics

▶ Watch it hereMore videos →

Key Takeaways

Want it done for you? See our local SEO services.

Google Analytics added a dedicated AI assistant channel on May 13, 2026, so ChatGPT and Gemini traffic can now appear as a named source in your reports. But the setup has gaps: Perplexity needs to be added manually, conversions must be marked before you measure them, and a large chunk of AI traffic still hides inside direct sessions. These steps close those gaps so you can see whether AI referrals are actually producing bookings.

• GA4’s AI assistant channel went live May 13, 2026, and groups ChatGPT and Gemini traffic automatically.
• Perplexity and smaller AI tools must be added by hand through the channel group editor in GA4 Admin.
• Always confirm your GA4 tag is firing before trusting any channel data.
• Mark your booking or inquiry event as a conversion in GA4 before you try to measure AI-referred outcomes.
• Build a custom Exploration report to compare AI versus organic conversion rates side by side.
• Direct traffic is not clean: AI referrals often land there when the referrer header gets stripped.
• Zero AI traffic in your reports is either a tracking gap or a recommendation gap, and each requires a different fix.

Why AI Referral Traffic Finally Has a Home in GA4

Google Analytics now tracks AI-referred sessions in a named channel called “AI assistants,” which went live on May 13, 2026, meaning you no longer need messy workarounds to separate ChatGPT and Gemini visits from the rest of your traffic. Before that date, those sessions scattered into direct, referral, or unknown, making it impossible to know whether an AI recommendation was actually producing any business. The May 2026 update is a first-party change inside GA4, confirmed by Book with Haven, and it applies to any property that already has a working GA4 tag.

Why does this matter for a home-service contractor or property manager? People are asking ChatGPT and Gemini questions like “find me a vacation rental manager near me” or “which HVAC companies in Asheville have good reviews.” When those tools recommend your business and a visitor clicks through, that session needs to land somewhere measurable in your reports. Without the AI assistant channel, the click just looks like a homepage visit with no origin, and you have no way to connect it to a booking form submission or a phone call.

The channel is a starting point, not a complete solution. Google groups the AI tools it recognizes automatically, but sources like Perplexity still require a manual rule. Newer tools may not be recognized at all. The practical approach is to treat the channel as a floor, the minimum you can see, and then verify every source that matters to your business individually. The steps below cover the full process from tag confirmation through conversion tracking and comparison reporting.

Confirming Your GA4 Tag Is Actually Working

Before checking any channel report, confirm your GA4 property is receiving live data, because no channel grouping works if the tag is broken or missing. Log into analytics.google.com, select your property in the top left corner, go to Reports, then Real-Time, and open your website in a separate browser tab. If your own session appears in the Real-Time view within about 30 seconds, the tag is firing correctly. If nothing appears, the tag is either absent from the page or blocked, and that issue must be resolved before anything else in this guide will produce reliable data.

How do I know if my GA4 tag is installed correctly?

The Real-Time report in GA4 is the fastest way to verify tag installation. Open your site in one tab and watch the Real-Time report in another. Your session should appear within 30 seconds. If it does not, check whether Google Tag Manager is published with the correct GA4 configuration tag, or look for a hardcoded gtag.js snippet in your page source and confirm the Measurement ID matches your property.

What breaks a GA4 tag without anyone noticing?

A tag can stop firing after a website update that removes the header scripts, after a caching plugin is configured to strip certain JavaScript files, or after a tag manager container is accidentally unpublished. Because most traffic reports still show historical data for days after a tag breaks, it is easy to miss the gap until you look at a date-specific range and notice sessions dropped to zero. Checking Real-Time once a week takes about two minutes and catches these breaks early.

Can I have more than one GA4 tag firing on the same page?

Yes, and duplicate tags are a common problem after site rebuilds. If two GA4 tags fire on a single page, sessions get double-counted, bounce rates appear artificially low, and channel groupings become unreliable. You can check for duplicates using the GA4 DebugView in Admin or by opening your browser’s network tab and searching for “collect” requests to see how many fire on a single page load.

Finding the AI Assistants Channel in Traffic Acquisition

Once your tag is confirmed, go to Reports, then Acquisition, then Traffic Acquisition inside GA4. The default table shows Session Default Channel Group as the first column. If any AI-referred sessions have arrived since May 13, 2026, you should see “AI assistants” listed as one of the rows in that table. The channel will not appear if no AI traffic has reached your site, or if the sessions are still scattering into direct or referral because the grouping rules were not yet active when those visits occurred.

Why is the AI assistants channel not showing up in my GA4 report?

The channel only appears in reports once at least one session has been grouped into it after the May 13, 2026 update date. If you see nothing in that row, either no AI tool has referred a visitor to your site in the selected date range, or the visits are being miscategorized as direct traffic because the referrer was stripped during the redirect chain. Try extending the date range to cover the full period since May 13 before concluding there is no AI traffic.

How do I see which specific AI tool sent each session?

Click on the AI assistants row in the Traffic Acquisition table, or add a secondary dimension called Session Source. This breaks the channel down by domain, so you can see chatopenai.com, gemini.google.com, perplexity.ai, and others listed separately. This level of detail matters because ChatGPT and Gemini can behave differently as referrers and may send visitors to different landing pages with different intent levels.

What is the difference between session source and session medium in GA4?

Session source is the specific domain or origin of the traffic, such as chatopenai.com or google.com. Session medium is the category, such as referral, organic, or none. For AI assistants, the source will show the AI tool’s domain and the medium will typically show referral. Looking at both together gives you the clearest picture of where a session originated and how it was categorized by GA4’s grouping rules.

Adding Perplexity to the AI Assistants Channel Manually

Perplexity does not fall into the AI assistants channel automatically, so sessions from that source will show up under referral or direct unless you add a custom rule. To fix this, go to Admin in the bottom left of GA4, find Data Display, and select Channel Groups. Open the Default Channel Group and click Edit. Locate the AI assistants channel in the rule list and open its conditions. Add a new condition: Session Source exactly matches perplexity.ai. Save the change. From that point forward, Perplexity sessions will group correctly into AI assistants.

One important limitation: this fix is forward-looking only. GA4 does not reprocess historical data when you edit channel group rules. Sessions that arrived before you made the change will stay in whatever bucket they landed in originally. This is why making the fix as early as possible matters. If you are currently seeing perplexity.ai appear as a referral source in your reports, those past sessions are already miscategorized, and the best you can do is note the date you made the correction and treat data before that date as incomplete for Perplexity specifically.

Are there other AI tools I need to add manually besides Perplexity?

Potentially yes. Any AI tool not explicitly recognized by Google’s default grouping rules will land outside the AI assistants channel. You can identify these by looking at your referral traffic and sorting by source domain. If you see domains like you.com, phind.com, or any AI-related source that is not in the AI assistants channel already, add each one using the same custom rule process you used for perplexity.ai.

Will editing channel groups affect my historical data?

No. GA4 channel group edits apply only to data collected after you save the change. Historical sessions are not reprocessed. This is different from how some older analytics platforms handled rule changes. The practical implication is that any comparison you make between periods before and after your edit will not be an apples-to-apples comparison for the sources you corrected. Document the date of every channel group change so you can account for it when reviewing trends over time.

Building a Custom Exploration Report for AI Referral Bookings

The standard Traffic Acquisition report gives you a channel-level view, but a custom Exploration in GA4 lets you slice the data by landing page, conversion event, and AI source all at once, which is what you actually need to evaluate whether ChatGPT or Gemini traffic is producing bookings. Go to Explore in the left navigation of GA4 and create a blank exploration. Name it something specific, like “AI Referral Bookings,” so it is easy to find later.

Add these dimensions to your exploration: Session Default Channel Group, Session Source, and Landing Page. Add these metrics: Sessions, Conversions, and the specific event name you use for a booking inquiry, such as form_submit or phone_click. Set a filter so the report only shows rows where Session Default Channel Group exactly matches “AI assistants.” Now you have a focused view: which AI tool sent the visitor, which page they landed on, and whether they completed a conversion action.

This setup also makes it straightforward to generate SEO reports in PushLeads alongside the GA4 data, so you can line up AI referral trends with your broader search visibility numbers in one workflow.

What dimensions should I include in an AI referral exploration in GA4?

At minimum, include Session Default Channel Group to confirm the filter is working, Session Source to see the specific AI tool, and Landing Page to see which URL the visitor arrived on. If you want to go deeper, add Device Category to see whether AI referrals come mostly from mobile or desktop, and add Date to track trends over time. Keep the dimension count manageable; too many dimensions in one exploration can make the table hard to read and slow to load.

How do I compare AI traffic conversion rates to organic search in GA4?

In your Exploration, use the Comparisons feature to create two segments: one filtered to Session Default Channel Group equals AI assistants, and one filtered to Session Default Channel Group equals Organic Search. Place both segments side by side in the same table. Watch Sessions and Conversions for each segment over the same date range. The conversion rate difference between AI-referred visitors and organic visitors tells you whether AI traffic arrives with higher or lower purchase intent.

Marking Conversion Events Before You Measure AI Traffic

Conversion tracking must be set up before you evaluate AI channel performance, because GA4 will not retroactively assign a conversion credit to past sessions once you mark an event. Go to Admin, then Events in GA4. Find the event that fires when a visitor completes your most important action, typically a contact form submission, a booking confirmation page view, or a phone number click. Toggle the “Mark as conversion” switch to on for that event. Now when you look at the AI assistants channel in any report, the Conversions column reflects actual business outcomes, not just sessions.

Which events should a home-service contractor mark as conversions in GA4?

Mark the event that represents real revenue intent. For most contractors, that is a contact form submission that triggers a thank-you page view, which you can track as a page_view event filtered to the thank-you URL. A phone number click tracked as a click event is also worth marking. Avoid marking soft actions like page scrolls or time-on-site thresholds as conversions; those inflate your conversion numbers without reflecting actual lead generation.

Can I add conversion tracking after my site has been running for months?

Yes, but only future sessions will be credited. GA4 does not backfill conversion data onto events that already fired before you toggled the switch. This means your conversion data for the AI assistants channel will only be meaningful starting from the date you marked the event. Set up conversion tracking as early as possible, even if you are not yet seeing AI traffic, so you have clean data from the moment visits start arriving.

What if my booking form is on a third-party platform?

If your booking form lives on a platform like Jobber, ServiceTitan, or a property management system, the thank-you page may be on a different domain, which breaks standard page-view based conversion tracking. In this case, configure a click event on the “Book Now” or “Submit” button on your own site and mark that as a conversion. It is not a confirmed booking, but it is the closest reliable signal you can track without server-side setup.

Reading the Numbers: What a Small Sample of AI Traffic Actually Tells You

A small number of AI-referred sessions does not mean AI traffic is unimportant. Consider a home-service business receiving around 200 sessions per month. If four of those sessions come from the AI assistants channel, that is 2% of total traffic. At a site-wide conversion rate of 5%, four sessions would produce roughly 0.2 conversions on average. That sounds negligible, but if those four AI-referred visitors convert at 10% because they arrived with specific, pre-qualified intent, that changes the math considerably.

This is exactly the kind of outcome the exploration comparison is designed to surface. In the video, I walk through a scenario where a local home-service company runs the full setup and after 30 days sees 11 sessions from chatopenai.com. Two of those 11 sessions submitted the contact form. That is an 18% conversion rate from ChatGPT-referred traffic compared to a 6% rate from paid search over the same period. That business did not know that before setting up the tracking. Now they do, and they can prioritize the work that improves their AI visibility accordingly.

Understanding what ChatGPT uses to recommend contractors is the other side of this analysis. The tracking shows you whether AI is sending traffic; the recommendation research shows you why some businesses appear and others do not.

How many AI-referred sessions do I need before the data is meaningful?

There is no universal threshold, but a practical minimum is around 30 sessions per source before conversion rate comparisons become reliable. Below that, a single additional conversion can swing your rate by several percentage points, making trends hard to interpret. If you are seeing fewer than 30 sessions from AI assistants in a 90-day window, focus first on improving your AI visibility so the sample size grows, then revisit the conversion rate analysis.

Should I weight AI traffic differently than organic or paid traffic?

Not necessarily, but pay attention to intent signals. AI-referred visitors often arrive after asking a specific question, which means they may have already compared options before clicking your link. That pre-qualification can show up as higher conversion rates, lower bounce rates, or higher average session duration compared to broader organic traffic. Watch those secondary metrics alongside conversions before drawing conclusions about traffic quality.

Common Mistake: Why Direct Traffic Is Not a Clean Bucket

Direct traffic in GA4 is not simply people who typed your URL into their browser. It is a catch-all for any session where GA4 could not determine an origin, and AI referrals fall into it regularly. When an AI tool opens your link in a new tab, the browser does not always pass a referrer header to your server. When a redirect chain is involved between the AI tool and your page, the referrer can be stripped entirely. The result is a session that looks like direct traffic in every report.

This means your AI assistant channel numbers are almost certainly an undercount. You cannot recover the misattributed historical sessions, but you can reduce future miscategorization in a couple of ways. First, make sure your landing pages load quickly without unnecessary redirects, since each redirect in the chain is another opportunity for the referrer to be dropped. Second, if you have any profile pages or citation listings inside AI tool ecosystems where links are allowed, add UTM parameters to those URLs so the source is embedded in the link itself rather than relying on the referrer header.

If you want to understand the full scope of why AI recommendations don’t convert at the rate you expect, the referrer stripping problem is part of the answer; the other part is what happens on the landing page after the visitor arrives.

How can I tell if AI traffic is hiding in my direct sessions?

Look at your direct traffic landing pages. If direct sessions are frequently landing on interior pages rather than your homepage, that is a signal some of those sessions have a referrer that was stripped. Direct traffic from people who typed your URL manually almost always lands on the homepage. Interior page landings within the direct bucket often indicate a referral that lost its source tag somewhere in the redirect chain, and AI tools are a common culprit given how they open links.

Do UTM parameters work for AI-referred traffic?

UTM parameters work any time you control the URL that gets shared. If you have submitted a business profile to an AI tool’s partner directory, or if you have placed a link in a resource that AI tools index, you can add UTM parameters to those links. When a visitor clicks a UTM-tagged link, GA4 reads the source from the URL itself and does not rely on the referrer header. The limitation is that you cannot add UTMs to links that AI tools generate dynamically from their own training data or live search results.

Common Mistake: Ignoring the Landing Page Column

Where AI traffic lands on your site matters as much as the fact that it arrived. If ChatGPT recommends your business and the visitor lands on your homepage with no clear next step, no booking form visible above the fold, and no phone number in the header, that lead is likely lost within seconds. The landing page column in your Exploration report tells you exactly where AI-referred visitors are arriving, and it is one of the most useful columns in the entire setup.

Once you know which pages AI traffic is hitting, audit those pages specifically. Does each one have a visible phone number? A booking form or a prominent link to one? A clear statement of what services you provide and where you provide them? AI tools often send visitors to blog posts, service area pages, or location-specific pages rather than the homepage, because those pages contain the specific information the AI used to recommend you. If the page that attracted the recommendation does not also convert the visit, you are losing leads that your AI visibility already earned.

This is also a good opportunity to review whether you should fix data gaps for AI visibility across your site, since the same structured content that improves conversions from AI-referred visitors also improves the likelihood that AI tools recommend you in the first place.

Which landing pages convert AI-referred visitors best?

Pages that directly answer the question the AI tool responded to tend to convert AI referrals better than general pages. If someone asked ChatGPT for a vacation rental manager in Asheville and clicked through to your Asheville property management services page, that page should immediately confirm you offer that service, show your geographic coverage, and present a low-friction next step like a phone number or a short inquiry form. Pages that require the visitor to navigate to find that information lose the visit before the conversion.

Should I create dedicated landing pages for AI referral traffic?

You do not need pages built exclusively for AI traffic, but you should ensure that every page AI tools are likely to cite meets a high conversion standard. Because AI tools tend to pull from service pages, FAQ pages, and location-specific content, those pages should be your priority for conversion optimization. If your exploration report shows AI visitors consistently landing on a blog post with no call to action, adding a booking form or a prominent contact link to that post is a straightforward fix.

What to Do When You See Zero AI Traffic

Zero AI traffic in your GA4 reports means one of two different things, and treating them as the same problem leads you in the wrong direction. The first possibility is a tracking gap: the AI assistant channel was not in place before May 13, 2026, your Perplexity rule has not been added yet, or some AI-referred sessions are hiding in direct or referral. The fix for a tracking gap is the setup process described in this post. Work through each step and check the data again after 30 days.

The second possibility is a recommendation gap: AI tools are simply not mentioning your business when people ask questions in your service category and location. No amount of GA4 configuration fixes that. The recommendation gap is a different problem that requires a different kind of work, specifically building the citations, structured data, reviews, and directory presence that AI tools pull from when constructing their answers.

If you want to know where you currently stand, one practical step is to test your brand across AI assistants on a monthly basis so you can see firsthand whether tools like ChatGPT and Gemini are mentioning you, and what they say when they do. That visibility check is separate from GA4 tracking but feeds directly into how you interpret what you see in your reports. If the check shows you are not being recommended, the goal shifts from “measure AI traffic” to “earn AI recommendations” first.

How long should I wait before concluding AI traffic is not reaching my site?

Give the full setup at least 30 days after confirming your tag is live, your Perplexity rule is added, and your conversion events are marked. AI referral traffic tends to be low volume early, especially for businesses that are not yet well-represented in the sources AI tools pull from. If after 60 days you still see zero sessions in the AI assistants channel, the more likely issue is that AI tools are not recommending your business yet rather than a GA4 configuration problem.

What sources do AI tools use to decide which businesses to recommend?

AI tools like ChatGPT and Gemini draw from a combination of web content they have indexed, structured business data from directories, review signals, and in some cases real-time search results. Having consistent NAP data across directories, strong review profiles on platforms like Google Business Profile and Yelp, and clear location-specific service pages on your website all contribute to your likelihood of being recommended. Learning more about how to get recommended by AI search gives you a fuller picture of which specific factors carry the most weight for local service businesses.

Frequently Asked Questions

When did Google Analytics add the AI assistants channel?

Google Analytics added the AI assistants channel on May 13, 2026, according to Book with Haven. Before that date, traffic from tools like ChatGPT and Gemini did not have a dedicated home in GA4 reports and scattered across direct, referral, or unknown categories. The change is a first-party update inside GA4, meaning it applies automatically to all GA4 properties without any action required from you, as long as your tag is working and your property was already set up before that date.

Does the AI assistants channel in GA4 track all AI tools automatically?

No. Google groups the major AI assistants it recognizes automatically, which currently includes ChatGPT and Gemini in most setups. Perplexity is noted by Book with Haven as requiring a manual rule addition. Other newer or smaller AI tools may also fall outside the automatic grouping. The practical approach is to check your referral traffic report for AI-related domains and add custom channel group rules for any source you care about that is not already grouped correctly.

Why does my GA4 show zero sessions in the AI assistants channel?

Zero sessions in the AI assistants channel can mean two things: either AI tools are not currently recommending your business and sending traffic, or the tracking setup has gaps such as a broken GA4 tag, missing Perplexity rule, or AI sessions being misattributed to direct traffic. Work through the full setup first, including tag verification, channel group rules, and a 30-day waiting period, before concluding that AI tools are not sending you any visitors at all.

How do I know if my direct traffic contains hidden AI referrals?

Check which landing pages your direct traffic sessions are arriving on. Direct sessions from people who manually typed your URL almost always land on the homepage. If a significant portion of your direct sessions are landing on interior pages like service pages, blog posts, or location pages, that pattern suggests some of those sessions have a stripped referrer, which is common with AI tool referrals. This is a signal that your AI traffic numbers are undercounted, though you cannot retroactively correct those sessions.

Do I need to set up conversion tracking before looking at AI traffic data?

Yes. GA4 does not retroactively assign conversion credit to past sessions once you mark an event. If you look at your AI assistants channel before marking any conversion events, the Conversions column will show zero regardless of how many bookings those sessions produced. Set up your conversion events as the first step after confirming your tag is working, so every AI-referred session captured after that point is associated with accurate conversion data from the start.

Can I use UTM parameters to track AI referral traffic more accurately?

You can use UTM parameters on any link you directly control that gets shared with AI-related audiences. This includes links in business profiles on AI tool partner directories, links in citation sources that AI tools index, and any resource pages where you can specify the URL format. When a visitor clicks a UTM-tagged link, GA4 reads the source from the URL parameters rather than relying on the referrer header, which makes the attribution more reliable. You cannot apply UTMs to links that AI tools generate autonomously from their training data.

How often should I check my AI referral traffic reports in GA4?

A monthly review is a reasonable cadence for most small businesses once the setup is complete. Because AI referral volumes tend to be low early on, weekly reviews of a small dataset can produce misleading trend impressions. At the monthly level, look at total sessions from the AI assistants channel, conversion rate compared to organic search, and the landing page breakdown. If you make changes to your website or AI visibility strategy, note the date so you can correlate those changes to shifts in the data.

What is the difference between a tracking gap and a recommendation gap?

A tracking gap means AI tools may be sending visitors to your site, but your GA4 setup is not capturing or categorizing those sessions correctly. The fix is technical: verify your tag, add channel group rules, and mark conversion events. A recommendation gap means AI tools are simply not mentioning your business when users ask relevant questions. The fix is strategic: build the citations, reviews, structured data, and content that AI tools pull from. Both gaps can produce zero AI traffic in your reports, but they require completely different responses.

Set Up Your AI Traffic Tracking This Week

The setup I have described here takes less than two hours to complete, and most of that time is spent waiting for the Real-Time report to confirm your tag. Confirm your GA4 tag is firing, find the AI assistants channel in Traffic Acquisition, add Perplexity manually, mark your booking event as a conversion, and build the exploration report with both segments. After 30 days, you will have real data showing whether ChatGPT and Gemini are sending you visitors, which pages those visitors land on, and whether they are converting at a rate that makes AI visibility worth your continued investment.

If you run through that process and see zero AI traffic after 60 days, the conversation shifts from tracking to visibility. That is where we at PushLeads spend most of our time with clients: building the citation profiles, review signals, and structured content that AI tools use when recommending local service businesses. If you want to know where you stand right now, reach out at pushleads.com/contact and we will start with a straightforward assessment of your current AI visibility before talking about anything else.

Related reading from PushLeads

• How to Prompt-Test Your Brand Across AI Assistants (Monthly Checklist)
• How to Generate SEO Reports in the PushLeads Dashboard
• ChatGPT Is Now Recommending Contractors. Here Is What Decides Whether It Is You.
• Why AI Search Ignores Your Contractor Business (Fix This)
• How to Get Recommended by AI for Local Services: The 44% Stranger Effect That Is Making Unknown Contractors the First Choice for Homeowners
• Why AI Recommendations Don’t Convert Local Service Leads: The 43% Verification Problem Costing You Booked Jobs

Your path · AI

Keep going: read these next

1. You’re here29 minHow to Track ChatGPT and Gemini Bookings in Google Analytics
2. Up next11 minThe AI Search Revolution: How 2 Billion People Are Finding Information in 2025
3. 49 minZero-Click SEO and AI Overviews: The Complete Guide to Winning When Users Never Visit Your Website
4. 45 minAI Search Optimization for Local Service Companies: Complete Implementation Guide
5. 47 minAI-SEO: How Artificial Intelligence is Revolutionizing Search Rankings in 2026

Up next

The AI Search Revolution: How 2 Billion People Are Finding Information in 2025

11 min read · AI search is reshaping how billions find information in 2025, fundamentally transforming the marketing landscape and requiring immediate strategy shifts.

Read now →▶ Watch Jeremy instead · 9:12

Want this done for you?

Scan your site. See the gaps. Decide with numbers.

• Rankings, Google Business Profile and site-speed teardown
• Report lands in your inbox in about a minute
• No obligation · 4.9 from 67 Google reviews

Your websiteRun my free audit

Prefer a human? Book 20 minutes with Jeremy

Continue reading→