Useful takeaways from this story.

68% of ads appeared in conversations with no sign of purchase intent, so many impressions likely reached non-buying moments.

Relevancy varied by sector: marketing/B2B and software/SaaS saw the highest unrelated rates, while data brokers, travel, and automotive had better contextual alignment.

# What the analysis looked at Searchable examined more than 11,000 ads served inside real ChatGPT conversations between 4 July and 4 August 2026. Each ad was paired with the conversation it appeared in and graded into three relevancy bands: direct match, contextual match, and unrelated.

# Core findings A third of ads (33%) were unrelated to the conversation where they appeared. Direct matches — ads explicitly advertising the product the user asked about — made up 27% of impressions. The largest single category, 40%, were contextual matches: ads linked to a topic mentioned earlier in the thread but not to the immediate question.

Two-thirds of served ads (68%) appeared in threads where the user showed no sign of wanting to buy, book, hire, or compare anything at any point. Counted by conversation rather than by individual ad, 40% of chats that carried ads had at least one unrelated ad, and in 28% of ad-bearing chats every ad was unrelated to the topic the user was discussing.

# Sector differences in relevancy

# How targeting works and why it misfires

# What OpenAI says and how the numbers compare OpenAI's then-chief revenue officer, Denise Dresser, said in June that user dismissals of ChatGPT ads have fallen by half since the ad business launched in February, using dismissals as a proxy for relevance. Searchable's month-long analysis, roughly six months into the rollout, found the share of ads with no connection to the conversation remained close to a third, with no clear improvement as ad volumes increased.

# Practical takeaway for advertisers and publishers Advertisers placing ads inside an assistant should expect a mix of direct, contextual, and unrelated placements. Many paid impressions are landing in non-buying moments. For brands that need high intent placements, an assistant ad placement is not a guaranteed substitute for being the actual answer in the conversation. Meanwhile, placements tied to life events or clear contextual signals (as seen in insurance and some travel cases) appear more likely to reach relevant moments.

# Bottom line ChatGPT ads produce a meaningful share of impressions that don't match the immediate conversation or any purchase intent. That gap is uneven across sectors and remains a factor for advertisers weighing reach inside conversational assistants against precision targeting.

More context around this story.

Loading more related stories...

Keep reading in the app

Open the app view to save this story, compare related coverage, and continue from the same source.

Open in app