A Third of ChatGPT Ads Appear in Irrelevant Conversations

ChatGPT ads are misfiring at an alarming rate. A new analysis of more than 11,000 ads served inside real ChatGPT conversations found that a full third of them had zero connection to what the user was actually discussing.

The findings come from Searchable, an AI visibility platform that paired every ad with the conversation it appeared in between July 4 and August 4, 2026. Each ad was graded on how closely the advertised product matched the user’s query. The results paint a messy picture for OpenAI’s young advertising business.

The Relevancy Breakdown

Just over a quarter of ads, 27 percent, were a direct match, meaning they advertised the specific product the user was asking about. The largest slice, 40 percent, landed in the contextual band. Those ads connected to something the user had raised earlier in the thread, but not to the question sitting in front of them.

That leaves the remaining third with no link to the conversation at all. No connection. No shared topic. Just an ad that showed up anyway.

No Purchase Intent Anywhere

The problem runs deeper than mismatched products. More than two-thirds of ads, 68 percent, appeared in conversations where the user showed no sign of wanting to buy, book, hire, or compare anything at any point in the thread. Advertisers are paying for impressions in chats where no purchase was ever likely.

Counting by conversation rather than by ad makes the issue even starker. It's brutal. Forty percent of ad-carrying chats contained at least one unrelated ad, and in 28 percent of those chats, every single ad served was off-topic, so the mismatch isn't a rare glitch but a systematic failure that undermines the whole premise of targeted advertising. But that's the reality we've uncovered.

An ad that lands in the wrong conversation is likely to be scrolled past, while the recommendation inside the response has already been shown to convert higher than other organic channels.

How ChatGPT Ads Actually Work

Traditional paid search flips the script. The advertiser picks the keywords that signal a buyer. ChatGPT’s ad platform works differently. It relies on context hints, where advertisers describe the conversations, situations, and topics they want to appear in, and OpenAI matches those against a live conversation.

This all takes place inside an AI engine that people also use to write, study, troubleshoot, and think through problems. That dual purpose creates a fundamental tension. The platform is trying to read intent from a stream of consciousness that often has no commercial angle at all.

Sector Misfire Rates

An advertiser's sector predicts how badly its ads misfire. But the numbers tell a stark story. Marketing and B2B services ads were unrelated in 47 percent of cases, a figure that stands as the worst rate measured, while software and SaaS ads came close behind at 45 percent, placing them second in a race nobody wants to win. That's it. Those two sectors bottom out the list.

Data brokers and background checks led the pack at 50 percent, a figure that still fell short of a direct relevancy match for more than half of all ads served. Travel hit 43 percent, while automotive trailed at 42 percent. That's the whole story. But here's the kicker: no sector managed to clear that 50 percent bar, so every single industry failed to achieve a majority of relevant ad placements, a sobering reality for marketers who've poured millions into precision targeting.

Insurance tells a different story. It was rarely unrelated at just 23 percent, but also rarely a direct match at 11 percent. Two-thirds of its ads fall in the contextual band, attached to a life event somewhere in the thread rather than the question being asked.

OpenAI’s Defense

Denise Dresser, then OpenAI’s chief revenue officer, said in June that the rate at which users dismiss ads inside ChatGPT has fallen by half since the advertising business launched in February. The company treats dismissals as its proxy for relevance.

The Searchable analysis covers a month roughly half a year into that rollout, and it's a snapshot that doesn't flatter the platform's early efforts. Across that period, the share of ads with no connection to the conversation stayed close to a third. That's stubborn. It didn't move. There was no sign of improvement even as ad volume grew, so the stagnation wasn't a matter of scale or testing conditions, it was a persistent pattern that held firm through increased pressure and activity alike. But the numbers never budged, not even slightly.

But does that really tell us the whole story? Dismissal rates might not capture what's actually happening. Users could be scrolling past bad ads without bothering to dismiss them, or they might be tuning out entirely, and that silence doesn't mean they're engaged,it could mean they've simply stopped caring. So the numbers alone can't answer it. They're just a snapshot, not the full picture.

What This Means for Brands

For brands, the consequence is clear. A placement inside an assistant is not a substitute for being the answer. An ad that lands in the wrong conversation gets ignored, while the organic recommendation inside the response has already been shown to convert higher than other channels.

ChatGPT ads are still young. The platform launched in February, and it's been scaling rapidly since, yet the data suggests relevance remains a serious hurdle for advertisers who are spending money on impressions that never had a chance to convert. But that's a costly problem. They can't afford to ignore it.

The takeaway for marketers is simple. Context hints are not keywords. The precision that made paid search work does not automatically carry over to an assistant that people use for everything from homework help to holiday planning. Until the matching improves, a third of that ad spend is going nowhere.