The honest answer is that this question can’t be answered fairly yet — not because the data doesn’t exist, but because Google has structured AI Overviews so that its referral traffic is nearly impossible to isolate, while ChatGPT’s traffic shows up cleanly in any standard analytics setup. So the real comparison in 2026 isn’t “which one sends more traffic” — it’s “which one you can actually prove is working.”
That distinction matters more than it sounds. A business owner deciding where to invest optimization effort needs to know not just which platform reaches more people, but which platform will let them measure the return on that effort at all.
Why Google’s side of this comparison is structurally unmeasurable
AI Overviews and Google’s AI Mode sit inside Google Search itself. When someone searches and gets an AI-generated answer at the top of the results page, any outbound click that follows gets logged in your analytics as ordinary “google / organic” traffic — the same bucket as a classic ten-blue-links click. Google doesn’t separately tag or report AI Overview referrals, and by design, a large share of AI Overview answers resolve the query without an outbound click at all. Reach estimates for AI Overviews and AI Mode combined put the monthly audience in the billions, which likely makes it the single largest AI surface in existence — and also the one you have the least visibility into.
This isn’t a temporary reporting gap that will get patched next quarter. It reflects how the feature is built: Google Search’s business model depends on keeping users inside Google’s own interface, and AI Overviews are a natural extension of that, not an exception to it. Expecting granular AI Overview referral data in GA4 any time soon is optimistic.
What you can actually measure: ChatGPT Search
ChatGPT, by contrast, behaves like a normal referring website when it sends someone to your site. That traffic arrives with its own referrer string, shows up as a distinct source in GA4, and can be segmented, tracked over time, and tied to conversions like any other channel. If you want to know whether your GEO work is paying off in visits and leads rather than just citations, ChatGPT referral traffic is the cleanest signal available today, precisely because it’s the one platform in this comparison that isn’t trying to keep the click for itself.
That doesn’t mean ChatGPT sends more total traffic than Google’s AI features — it very likely doesn’t, given the sheer scale difference in reach. It means ChatGPT is the one you can actually put a number on, which is a very different and more useful thing when you’re deciding where to spend your time.
So which platform should a small business actually optimize for?
Both, but for different reasons and with different expectations attached. Optimizing for Google’s AI surfaces is worth doing because of sheer reach — being cited or featured in an AI Overview puts your business in front of an enormous number of searches, even if you’ll never get a clean number showing exactly how many of those impressions turned into visits. Treat that work the way you’d treat classic brand SEO: valuable, real, and only partially measurable, with success showing up indirectly through gradually rising direct and branded search traffic over time.
Optimizing for ChatGPT is worth doing because you can actually close the loop: publish content structured to be citation-worthy, watch referral traffic and conversions from that specific source, and iterate based on what’s actually landing rather than what you hope is landing. For a small business trying to prove that GEO spend is worth the ongoing investment, that measurability is genuinely valuable — it’s the difference between reporting “we think this helped” and reporting an actual number.
Side by side
| ChatGPT Search | Google AI Overviews / AI Mode | |
|---|---|---|
| Referral traffic in analytics | Visible and attributable, like any other channel | Bundled into organic search — not separable |
| Estimated reach | Smaller, but growing | Vastly larger — reportedly billions of monthly users |
| Outbound clicks | Sent to the source site when it answers from the web | Often suppressed by design; many queries resolve with no click |
| How to judge success | Direct: referral sessions, conversions from that source | Indirect: citation presence, gradual lift in branded/direct search |
A practical framework: measure what you can, hedge what you can’t
Rather than picking a side, the more useful approach for a small business is to treat these as two different jobs. Set up ChatGPT (and other AI assistants that send identifiable referral traffic) as a tracked channel in your analytics, the same way you’d track a new paid channel — with its own segment, its own conversion tracking, and its own review cadence. For Google’s AI Overviews, accept that you’re optimizing on faith to a real degree, and use secondary signals instead of direct attribution: track whether you appear in AI Overviews for your priority queries using a rank-tracking or SERP-monitoring tool, watch branded search volume over time, and treat a slow lift there as your best available proxy for “this is working,” even without a click count to point to.
The mistake to avoid is assuming that because Google’s AI traffic isn’t measurable, it isn’t happening — or conversely, that because ChatGPT’s traffic is measurable, it’s automatically the more important platform to chase. The measurability gap is an artifact of how each platform is built, not a signal of which one actually matters more to your business.
What this means for how you write content
If you’re producing content with both platforms in mind, the practical implication is that you’re writing for two different kinds of success at once, and it’s worth being explicit about which one a given piece is aimed at. Content built to earn an AI Overview citation should prioritize being unambiguous and quotable on the exact phrasing of common questions in your niche — the kind of sentence a model can lift cleanly and attribute without needing to interpret nuance. Content aimed at earning a ChatGPT click needs something extra: a genuine reason to visit the source rather than just read the summary, whether that’s a calculator, a comparison table, a downloadable template, or specificity that a short AI-generated answer can’t fully capture. Writing exclusively for the first goal tends to produce content that gets cited everywhere and visited nowhere, which is a fine outcome for brand visibility but a frustrating one if traffic and leads are the actual goal.
It’s also worth checking your own server logs occasionally, not just GA4. AI crawlers and assistants that fetch pages to answer a live query sometimes show up in raw access logs before — or instead of — generating a referral visit, and that pattern can tell you which of your pages are actually being pulled into AI answers even when no session ever gets attributed to it.
Will Google ever break out AI Overview traffic separately?
There’s no indication that’s coming soon, and the structural incentive points the other way — keeping the click inside Google’s own results page is central to how AI Overviews were designed, not an oversight in the reporting. Businesses shouldn’t plan around a fix arriving; the more productive move is building measurement habits that work around the gap rather than waiting for it to close.
Does a citation in an AI Overview matter if it doesn’t send a click?
It still matters for the same reason a strong brand SEO position has always mattered even when a searcher doesn’t click through: repeated exposure at the top of a search result builds recognition and trust, which shows up later as branded search, direct traffic, and word of mouth — just not in a single attributable session you can point to.
For more on how these AI search surfaces compare to each other, Google AI Mode vs. AI Overviews in 2026 breaks down how those two Google features relate to one another. If you’re still deciding whether GEO deserves a place in your marketing budget at all, is GEO worth it in 2026? lays out that case directly. And once you have some AI referral traffic to look at, how to tell if your AI search visibility is actually making you money covers how to turn that traffic into a real answer about ROI.