News — On August 10, 2026, SE Ranking published an analysis of 50,006 commercial prompts, collected on July 23, 2026 across 20 industries, showing that ChatGPT displays a sponsored placement on 25.94% of commercial answers, while the exact advertised URL appears as a cited source in only 0.09% of cases (SE Ranking, August 10, 2026).
Buying an ad in ChatGPT does not put you inside the answer. It puts you next to it. That is the most counter-intuitive finding in the study published on August 10, 2026 by SE Ranking, authored by Yulia Deda, which ran 50,006 commercial prompts across 20 industries.
The headline number is 25.94%: more than one commercial answer in four now carries a sponsored placement. The number that actually matters is 0.09%.
Three numbers worth keeping
Looking closer, only 3.63% of advertiser domains show up anywhere in the cited sources, and the advertiser brand is mentioned in just 4.44% of answers. In other words, 96.37% of ad placements come with no organic citation at all.
Why that gap changes everything
In a classic search engine, the ad and the organic result compete for the same visual space and the same attention. In a conversational interface, the logic differs: the answer text is generated from the sources the model retrieved, while the ad is pinned underneath, in a separate labelled block.
These two mechanisms do not talk to each other. Paying feeds the sponsored block. Being cited feeds the answer itself, the one the user reads and bases a decision on. It is the same conclusion we reached when analysing the structure of referral traffic coming out of ChatGPT: the value sits in the body of the answer, not in its margins.
The 14.35% off-topic ad rate compounds the problem. SE Ranking reports wide variation by industry, from 2.6% in pets to 54.2% in news and politics. In some segments, more than one ad in two fails to match the intent of the question.
ChatGPT and Google AI Mode make different calls
The comparison with Google AI Mode, run inside the same study, exposes two opposing philosophies on sensitive topics.
| Category | ChatGPT | Google AI Mode |
|---|---|---|
| All commercial queries | 25.94% | 29.45% |
| Healthcare | 28.69% | 2.64% |
| News and politics | 28.76% | 6.80% |
On overall volume, Google actually monetises slightly more. But in healthcare the gap is tenfold: Google all but bans advertising on YMYL queries, where ChatGPT applies its usual rate. For a healthcare, legal or financial brand, that divergence dictates two different strategies across the two platforms.
Key point: in ChatGPT, advertising and citation are two separate channels that do not substitute for each other. A media budget will not offset a missing citation, and a citation cannot be bought.
What this changes for a small or mid-sized business
Three operational consequences follow.
Conversational advertising is an acquisition channel, not an authority channel. It can generate a click. It will not make you the reference the model cites when a prospect asks which solution to pick. Those two goals need two budgets and two methods.
Organic space tightens visually, not structurally. A quarter of commercial answers now carries a sponsored block pushing everything else down. Cited sources remain the backbone of the answer. That matches what the GEOsurge study on model preference for established brands showed: citation rewards consistency and track record, not bidding.
You have to measure both separately. Tracking ad impressions says nothing about your presence in the answers themselves. The Search Console AI visibility report and citation-tracking tools cover the second question, the one no ad platform reports on.
What to do now
- Run your twenty commercial queries through ChatGPT and Google AI Mode, and record for each whether you are cited, mentioned, or absent. That is your baseline, and it takes an hour.
- Split the two budget lines. If your 2026 plan treats AI visibility as a sub-item of paid media, the 0.09% figure says you are funding the wrong lever.
- Handle healthcare, legal and finance separately. The 28.69% versus 2.64% gap means a competitor can buy space in ChatGPT where Google refuses to sell it.
- Make your pages retrievable. The trust signals models use to select their sources remain the only route to a citation.
The limits of this study
Three caveats are worth stating. The data was collected on July 23, 2026, on a single day: advertising inside ChatGPT is moving fast, and coverage rates have already shifted several times since rollout began. The sample covers English-language commercial prompts, mostly on the US market, which says nothing definitive about rates observed in Europe. Finally, the study measures citation presence, not clicks or conversions: it does not claim a ChatGPT ad is ineffective, it shows that an ad does not buy a citation.
Context matters too: OpenAI announced on January 16, 2026 that it was beginning to test advertising in ChatGPT in the United States, limited to free accounts and the Go tier, according to CNBC. The system is less than eight months old.
The Cicero take
That 0.09% is the best news of the summer for companies producing serious content. It proves the seat inside the answer is not for sale. As long as citation stays earned rather than bought, a mid-sized player with precise, sourced content can hold ground no media budget will take.
The question to ask is not what a ChatGPT ad costs, but who the model cites when a prospect asks the question. No ad platform reports that answer.
Sources
- → SE Ranking : study published August 10, 2026 by Yulia Deda, 50,006 commercial prompts across 20 industries, data collected July 23, 2026.
- → Search Engine Land : coverage and analysis of the study, August 10, 2026.
- → CNBC : OpenAI begins testing ads in ChatGPT in the US, January 16, 2026.
Growth and SEO content strategist, I founded Cicéro to help businesses build lasting organic visibility — on Google and in AI-generated answers alike. Every piece of content we produce is designed to convert, not just to exist.
LinkedIn