ChatGPT ads have been one of the most talked-about new placements in digital marketing — a chance to reach people inside the AI answers they increasingly trust. But talk is cheap, and until now there has been very little hard evidence on what a ChatGPT ad visit is actually worth compared with the Google Search ads most advertisers already run.

That evidence has now arrived, and it is sobering. A working paper from researchers at WU Vienna and the Frankfurt School, reported by Search Engine Land on November 2, analysed around 10 billion sessions across 1,898 websites and found that paid ChatGPT visits generated 64% less revenue per session than Google Search ad visits.

Before anyone rewrites their media plan, though, the study deserves a careful read — including its limits. Here is what it found, what it does not prove, and how advertisers should respond.

Key Takeaways

• Paid ChatGPT visits were worth $0.89 per session, against $2.45 for Google Search ads — a 64% gap.
• ChatGPT ad traffic converted at 0.37%, versus 1.17% for Google Search ads.
• 97% of the 87 advertisers running both channels earned less per ChatGPT visit.
• Researchers estimate ChatGPT ads would need to cost about $0.26 per click to match Google’s return on ad spend — against one disclosed price of $1.72 per click.
• This is an observational study of an early market, not a causal experiment — early evidence to weigh, not a verdict on the channel.

What Did the Study Find?

The headline finding is that paid visits from ChatGPT ads were worth $0.89 per session, compared with $2.45 per session for Google Search ads — meaning each ChatGPT visit generated roughly 64% less revenue. Conversion rates showed the same pattern, at 0.37% for ChatGPT ads versus 1.17% for Google Search ads.

The gap was not driven by a few unlucky advertisers. Among the 87 advertisers in the data who ran both ChatGPT ads and Google Search ads, 97% earned less per ChatGPT visit than per Google visit. When a result holds for nearly every advertiser measured, it points to something structural about the channel today — the audience, the intent, the formats, or the pricing — rather than individual campaign mistakes.

The study also looked at what happened to organic referral traffic from ChatGPT over the period, and that part of the findings needs much more careful handling, which we cover in the limits section below.

How Was the Study Conducted?

The research is an October 2026 working paper from academics at WU Vienna and the Frankfurt School, based on roughly 10 billion sessions across 1,898 websites between June 2025 and August 2026. It compares the revenue and conversion behaviour of paid ChatGPT visits against Google Search ad visits over that period.

That scale is the study’s strength: this is not a survey or a handful of case studies, but observed behaviour across billions of real sessions and nearly two thousand sites. It is also worth being precise about what kind of evidence it is. The researchers observed traffic and outcomes as they occurred; they did not randomly assign advertisers or users to channels. That makes it an observational study — powerful for describing what happened, weaker for proving exactly why it happened. We return to that distinction below, because it decides how much weight the numbers should carry in a budget meeting.

How Big Is the Gap Between ChatGPT Ads and Google Search Ads?

On the two measures advertisers care about most — value per visit and conversion rate — the gap is wide. A paid ChatGPT session was worth $0.89 against $2.45 for a Google Search ad session, and ChatGPT traffic converted at 0.37% against Google’s 1.17%, so a Google Search visit was roughly three times as likely to convert.

Put together, those two figures explain the headline. Advertisers are not just getting fewer conversions from ChatGPT; each visit is worth less when it lands. For a channel sold on the promise of reaching high-intent users inside AI answers, that is an uncomfortable result — users arriving from a conversation appear, on this evidence, to be earlier in their journey than users arriving from a search results page, where the query itself often signals readiness to buy.

But a gap measured today is not a gap guaranteed forever. ChatGPT ads are a new format in an immature market, and the comparison is against the most refined advertising machine in digital marketing. The right question is not whether ChatGPT ads trail Google now — on this evidence, they do — but whether the gap reflects the channel’s ceiling or simply its starting point.

What Would ChatGPT Ads Need to Cost to Match Google’s Returns?

The researchers estimate that, at current performance levels, ChatGPT ads would need to cost around $0.26 per click to deliver the same return on ad spend as Google Search ads. That estimate sits awkwardly against the one actual price disclosed in the research: $1.72 per click.

That spread — paying $1.72 for clicks that would need to cost $0.26 to break even on Google’s terms — is the commercial heart of the study. It suggests that, at the prices at least some advertisers are paying, ChatGPT ads are priced for the value advertisers hope the channel will have rather than the value it currently delivers. Pricing in new channels often starts optimistic and corrects as performance data accumulates; alternatively, performance itself may improve as formats and targeting mature, closing the gap from the other side.

Either way, the message for buyers is practical: if you are testing ChatGPT ads at premium click prices, the burden of proof is on the placement, and your own per-visit numbers — not the platform’s pitch — should decide whether the test continues.

What Are the Limits of This Study?

Three limits matter. First, it is observational, not causal: it shows ChatGPT visits were worth less, not that the ads themselves caused the difference. Second, the reported organic-referral decline of 5.9% was not statistically significant. Third, the higher 7.2–8.4% decline figures appear only under the study’s modelling assumptions.

Each of those limits changes how the findings should be used. Because the study is observational, differences in who clicks, what they were researching, and which advertisers chose to test the channel are all baked into the numbers. Because the unadjusted organic decline did not reach statistical significance, no one should quote a precise organic traffic fall from this paper as established fact — and the 7.2–8.4% range should always be described as model-dependent, never as the measured result.

Finally, this is a snapshot of an early market. ChatGPT’s ad formats and targeting are immature, and the users clicking today may not resemble the audience of a mature product. The study is best read as a rigorous baseline — the starting point against which future improvements should be measured — rather than a final judgement on AI advertising.

Why Might Early ChatGPT Ads Underperform?

The most plausible explanation is intent. People use ChatGPT to research, compare, and think — often long before they are ready to buy — while a Google search query is frequently typed at the moment of decision. An ad shown inside a research conversation meets a different mindset than an ad answering an active search.

Format maturity plays a part too. Google Search ads are the product of more than two decades of refinement in matching, formats, and measurement. ChatGPT ads are new: the formats are still developing, targeting is comparatively blunt, and advertisers are still learning which messages work inside a conversational interface. Early channels almost always look inefficient next to mature ones, because the mature channel’s numbers already include years of optimisation by thousands of advertisers.

None of that erases the study’s findings — advertisers spending money today are entitled to judge today’s returns. It does mean the findings describe a young channel’s current performance, not its permanent potential, and budget decisions should be written in pencil rather than carved in stone.

How Should Advertisers Test ChatGPT Ads?

Test ChatGPT ads the way you would test any unproven channel: small, capped budgets, judged on your own per-visit value and conversion data, compared like-for-like against your existing channels. Set the success threshold before you spend, and let the numbers — not the novelty — decide whether the test scales.

In practice, that means:

• Start with a ring-fenced test budget that cannot hurt your core campaigns if it returns nothing. Treat it as research spend, not performance spend.
• Measure value per visit, not just clicks or CTR. This study’s central lesson is that a click is not a click: track revenue per session and conversion rate for ChatGPT traffic separately in your analytics from day one.
• Compare like-for-like. Judge ChatGPT ads against the same products, landing pages, and time periods as your Google campaigns — and if you want help structuring campaigns and measurement properly, that is exactly the kind of work covered by our digital marketing services.
• Allow for different intent. A research-stage audience may need different creative and landing pages than search traffic. Test messaging built for consideration before concluding the channel cannot convert.
• Review against your threshold, then decide. If per-visit value approaches your break-even — the study’s $0.26-per-click estimate is a useful benchmark for what “competitive” would need to look like — scale cautiously. If it does not, pause and revisit as the format matures.

What Does This Mean for Your Ad Budget Right Now?

For most advertisers, the sensible response is to keep Google Search as the proven core of paid acquisition while treating ChatGPT ads as a measured experiment. The study gives no reason to abandon testing — and no justification for shifting serious budget into ChatGPT ads at current prices and performance.

The wider context matters too. AI surfaces are becoming advertising surfaces everywhere at once: as we covered in our breakdown of Google’s privacy policy change and what it means for advertisers, the data that informs targeting inside AI products is expanding, and placements inside AI answers will keep multiplying. Advertisers who learn how these channels perform now — cheaply, with proper measurement — will be ahead of competitors who either ignore them or overpay for them. And because AI answers increasingly shape which brands get considered at all, paid testing pairs naturally with organic AI visibility work such as getting your business cited by ChatGPT and Perplexity.

Frequently Asked Questions

How much less revenue do ChatGPT ads generate per visit than Google Search ads?

According to the study, paid ChatGPT visits were worth $0.89 per session compared with $2.45 for Google Search ads — about 64% less revenue per visit.

How many advertisers earned less from ChatGPT ads in the study?

97% of the 87 advertisers who ran both ChatGPT ads and Google Search ads earned less per ChatGPT visit than per Google Search visit.

What conversion rates did the study report for ChatGPT ads vs Google Search ads?

ChatGPT ad traffic converted at 0.37%, compared with 1.17% for Google Search ads — meaning Google Search visits were roughly three times as likely to convert.

Does this study prove ChatGPT ads don’t work?

No. It is an observational study of a new, immature ad channel, not a causal experiment. It shows ChatGPT visits were worth less during the period studied; it does not prove the channel cannot work as formats, targeting, and pricing mature.

What price would ChatGPT ads need to match Google’s return on ad spend?

The researchers estimate ChatGPT ads would need to cost about $0.26 per click to match Google’s return on ad spend at current performance — against one disclosed price of $1.72 per click.

Should advertisers stop testing ChatGPT ads?

No — but tests should be small and judged rigorously. Use capped budgets, measure revenue per visit and conversion rate separately, compare like-for-like with your Google campaigns, and scale only if your own numbers clear a threshold you set in advance.

Final Thoughts

The first rigorous numbers on ChatGPT ads are a useful reality check: at today’s prices and performance, a ChatGPT visit is worth far less than a Google Search visit, and nearly every advertiser running both earned less from it. But the study is a baseline, not an obituary — an observational snapshot of an immature channel whose formats, targeting, and pricing are all still moving. Advertisers who test carefully, measure per-visit value honestly, and refuse to pay tomorrow’s prices for today’s performance will be positioned whichever way the gap moves next.