83% of ChatGPT Ad Triggers Slip Past Google Shopping

ChatGPT ad triggers are exposing a monetization gap that Google Shopping was not built to capture: 83% of the queries that trigger ads inside OpenAI’s ChatGPT would not have activated a Google Shopping ad.

That finding, from Similarweb and reported by PYMNTS, points to something bigger than a new ad format. It suggests AI assistants can identify commercial intent before a user has turned that intent into a clean product query.

Similarweb found that 83% of ChatGPT ad-triggering queries would never have activated a Google Shopping ad.

That is the thesis marketers need to sit with. ChatGPT ads are not simply competing for the same final click that search has long monetized. They are reaching users while they are still forming the buying question.

ChatGPT’s 83% ad-trigger gap exposes Google Shopping’s blind spot

Google Shopping is strongest when intent is already explicit. A user names a product, category, brand, or purchase-oriented phrase. The system can match that declared demand to a product ad.

ChatGPT, by contrast, can surface ads inside a longer exchange where the user may not begin with commercial intent at all. PYMNTS describes the difference clearly: Google Shopping matches ads to declared product queries, while ChatGPT can surface ads during problem-solving conversations, research sessions, and multi-turn exchanges.

That matters because Similarweb found that 46% of users who start a ChatGPT session with no commercial intent develop buying signals before the conversation ends. In other words, the ad opportunity may appear after the assistant has helped the user clarify the problem.

For advertisers, the tension is sharp. Search ads monetize people who already know how to ask for a product. ChatGPT ad triggers can reach people while they are still defining needs, constraints, and options. That could make the channel more valuable in some journeys, but it also makes performance harder to measure with old search habits.

This follows a broader shift in enterprise use of AI systems, where the quality of retrieval and recommendation matters more than simple access to a model. As we covered in Bad LLM Platforms Break Enterprise Knowledge Search, weak AI search can create bad decisions fast. The same risk applies when paid placements enter the answer flow.

The Similarweb data shows a new intent map, not proven ad superiority

The 83% figure does not prove that ChatGPT ads convert better than Google Shopping ads. It does not prove lower customer acquisition costs. It does not prove higher click quality.

It proves something narrower, but still important: most ChatGPT ad-triggering prompts do not resemble the query patterns that traditionally power Google Shopping placements.

That distinction should keep marketers from overreacting. A new surface for intent is not automatically a better performance channel. PYMNTS notes that benchmarks for click quality and conversion rates in a conversational environment are still early, even though OpenAI has added tools that look familiar to performance marketers.

Those tools now include cost-per-click bidding, a Conversions API, and pixel-based measurement. In May, OpenAI also launched a self-serve Ads Manager in beta and removed the minimum spend requirement that had previously limited access to larger commitments.

The ad product is moving toward the Google and Meta playbook. The user environment is not.

XOOMAR analysis: the real test is not whether ChatGPT can find commercial intent. Similarweb’s data says it can. The test is whether advertisers can separate incremental demand from interest that would have converted through search anyway.

Turn 15 is where ChatGPT ads get interesting

The most revealing part of the Similarweb data is not just the 83% gap. It is where ads appear inside the conversation.

Three weeks before a separate June Similarweb report, leading advertisers were seeing ads fire at a median of turns 7 to 8. By June 8, that had shifted to turns 14 to 22 for the same brands. Vanta’s average ad-firing turn reached 15.8. Monday.com’s reached 13.6.

That is a very different commercial moment from a standard search result. By turn 15, a user may have already compared options, asked follow-up questions, and narrowed a shortlist before seeing a sponsored placement.

PYMNTS also reports that 41% of ad moments inside ChatGPT are purely research-oriented. That creates both an opportunity and a trust problem.

The opportunity is obvious: an ad can appear when the user is more informed than they were at the first prompt. The risk is just as clear: if a sponsored recommendation feels like it is steering advice rather than helping the task, users may treat it as contamination of the assistant experience.

This is where OpenAI has a narrow path. It crossed $100 million in annualized ad revenue within six weeks of launching its U.S. advertising pilot, according to Reuters as cited by PYMNTS, while fewer than 20% of eligible U.S. users were being shown ads daily at that point. The revenue signal is strong. The user trust test is still underway.

Monday.com’s lead shows early ChatGPT ads skew toward B2B software

Monday.com holds 5.05% of global ChatGPT ad impressions, according to Similarweb data cited by PYMNTS. The top five advertisers collectively account for nearly 17% of impressions.

The early category pattern is also telling: eight of the top 10 advertisers by impression share are B2B SaaS or productivity brands. That fits the current evidence better than broad claims about every consumer category moving into AI ads.

B2B software is a natural early proving ground because users often research workflows, compare tools, and refine requirements before they are ready for a vendor page. That is exactly the kind of journey ChatGPT can observe across multiple turns.

Similarweb also noted that brands entering a newly launched market on the platform appear to have roughly 30 days of category-level advantage before competition intensifies and auction dynamics normalize.

That advantage will not last if the market behaves like other auction-based ad systems. But for now, early movers may be buying learning, not just impressions. They get to see which prompts, contexts, and late-stage conversation patterns produce action.

For companies building internal AI workflows, that lesson rhymes with a different problem: teams need useful AI systems that fit the task, not generic model access. Our guide to cutting remote work chaos with an AI productivity stack makes the same point from the buyer side. Context drives usefulness.

Advertisers need proof before shifting budget from search

The strongest counterpoint to the ChatGPT ad thesis is simple: Google Shopping intent is explicit, measurable, and familiar. ChatGPT intent is inferred, conversational, and still early.

Advertisers will not move major budgets just because a new format finds queries Google Shopping misses. They will demand evidence that those impressions drive incremental conversions, not softer consideration that would have happened anyway.

That means marketers should not treat ChatGPT ad triggers as keyword replacements. They should treat them as a new intent layer.

Practical tests should focus on:

• Prompt context: Which conversation patterns produce ad exposure?
• Turn depth: Do later ad placements perform differently from earlier ones?
• Downstream action: Do users click, convert, or return later through another channel?
• Incrementality: Would the sale have happened through search or direct traffic anyway?
• Disclosure quality: Does the sponsored placement feel clearly labeled and useful?

XOOMAR analysis: the winning creative will likely look less like a search headline and more like a decision aid. But that remains an inference from the structure of the format, not a proven performance rule from the available data.

The next test is whether useful ads can stay visibly paid

The future of ChatGPT ad triggers will come down to trust, measurement, and timing.

If ads keep appearing after deep research turns, they may become valuable for complex buying decisions where users need comparison and explanation before they act. The current data already shows late-turn placement, concentrated B2B advertiser share, and commercial signals emerging from initially non-commercial sessions.

If conversion quality disappoints, the 83% gap will look less like a new market and more like noisy intent. If users feel paid influence is blurring into advice, OpenAI will have to pull back or redesign the experience.

The evidence that would confirm the bullish case is straightforward: stronger conversion benchmarks, clear incrementality, stable user engagement, and transparent labeling that does not break trust. The evidence that would weaken it is just as clear: high impressions, low action, advertiser churn, or user resistance to sponsored recommendations inside AI answers.

For now, the signal is real but unfinished. ChatGPT ads have found commercial intent that Google Shopping would miss. The hard part is proving that intent is worth more than curiosity.

The Bottom Line

• AI assistants may create ad opportunities before users know exactly what they want to buy.
• Google Shopping could miss a large share of commercial intent emerging in conversational AI.
• Advertisers may need new strategies for targeting multi-turn research and problem-solving sessions.