Can a UK business pay to appear in ChatGPT answers, and is that the same as being recommended
There are two different things a buyer might mean by this, one of them is purchasable and one is not, and a British advertiser has an extra obligation attached to the first.
Published by AI Knows Us · Updated 22 October 2026
A British business asking whether it can pay to appear inside ChatGPT answers is usually asking one of two questions, and they have opposite answers. Advertising placements inside and around assistant products are a real and growing commercial channel. Being named in the body of an answer as a recommendation is not something anybody sells. Mixing them up is how budgets get committed to the wrong thing.
The two things, kept apart
| A paid placement | Being named in the answer | |
|---|---|---|
| What it is | An advertisement served by the operator, in or around the product | Your business appearing in the answer text as a suggestion |
| Can you buy it | Yes, through the operator's advertising products where they exist in your market | No. There is no placement to buy |
| What it depends on | Budget, targeting and the operator's policies | What material about you exists in the sources the engine reads |
| What happens when you stop paying | It stops that day | Nothing, until the sources change |
Anybody offering to place your business inside the organic answer text for a fee is selling something they do not control. The honest version of that offer is publishing work, which changes what the engine reads, and it should be priced and described as publishing work.
The British obligation attached to the paid version
If you pay for a placement, it is an advertisement, and United Kingdom advertising rules require advertising to be obviously identifiable as such. That principle applies to the format whatever it is called, and it applies to you as the advertiser as well as to the platform. Before you commit budget to a new placement format, confirm with your agency or your solicitor how the identification requirement is satisfied in that format, and keep a note of the answer.
The same applies in the other direction to anything you pay somebody else to publish about you, including creator content and sponsored articles. A payment makes it an ad regardless of where it sits.
What the operator documentation tells you about the machinery
One piece of evidence worth having in this conversation, because it is checkable by anyone. OpenAI documents four separate crawler identities rather than three: GPTBot, OAI-SearchBot, ChatGPT-User and OAI-AdsBot. Read from the operator's own bot documentation on 16 October 2026. The fourth exists specifically for the advertising side.
We then fetched the robots.txt file of 26 domains our own audits had named, following redirects, and parsed each one into per agent rules. 24 of the 26 were readable. Not one of the 24 blocked GPTBot, OAI-SearchBot or OAI-AdsBot at the root. And not one of the 24 named OAI-AdsBot at all, while 9 named GPTBot and 9 named PerplexityBot.
That is a useful fact for a buyer. The advertising identity is in the operator's documentation and in nobody's site configuration, which tells you how new this part of the market is. If you are planning around it, plan on the basis that the rules and the formats will move, and keep the date on any note you write about them.
What the organic half actually responds to, measured in Britain
On 22 October 2026 we put a 198 question set of British buying questions to ChatGPT and Gemini, with the market fixed inside the prompt because the gl, hl and location parameters on the search API are accepted, return 200 and answer from somewhere else. That gave 64 blind readings, 32 per engine.
- 30 of the 64 named no supplier at all. There is no incumbent in half of these answers, and no amount of advertising budget changes who the answer names.
- A single British agency website was cited in 19 of the 64 readings, and in 18 of ChatGPT's 32. One company's own pages, read constantly, with no placement involved.
- That same site was named in the answer text of only 4 of those ChatGPT readings. Being read and being recommended are two different outcomes and this is the clearest example we have measured.
- The engines are not comparable on sourcing. ChatGPT carried 2,656 of the 2,747 citation URLs in the set. Gemini carried 91, and 7 of its 32 readings carried none at all.
So the thing that puts a British business into the organic half of an answer is material the engines read, and the thing that gets it named is material that answers the recommendation question specifically. Those are two pieces of work and neither is a media buy.
How to decide between the two with a real budget
- Take a reading first. Ask your buyer questions and record who is named and which sources are cited. If nobody is named in your category, the organic half is unusually cheap to win and advertising is the expensive route to the same visibility.
- If you advertise, treat it as advertising. Measure it on cost per enquiry like any other placement, and satisfy the identification requirement.
- Do not buy a placement that claims to affect the answer text. There is no such product and the proposal is a signal about the supplier.
- Keep the two reported separately, because they will behave completely differently when you pause spend.
What British businesses ask about paying for AI visibility
Is there an ads product in ChatGPT today in the UK? Advertising formats in assistant products are being rolled out and the availability and the format differ by market and change quickly. Check the operator's own current advertising documentation rather than a summary, and note the date you read it.
Can an agency get us mentioned for a fee? It can do publishing work that changes what the engines read, which is legitimate and should be described as such. It cannot buy a mention in the answer text.
Does paying for ads help the organic mentions? Treat them as separate systems unless you have measured otherwise on your own data. Assuming a spillover is how an advertising budget gets justified with an organic result.
Do we have to disclose a paid placement? UK rules require advertising to be identifiable, and how that is satisfied in a given format is the question to put to your agency or solicitor before the campaign rather than after.
What is the cheapest first step? One reading. It tells you whether your category's answers name anybody, and that single fact decides whether advertising or publishing is the better use of the money.
The measurement that keeps the two honest
Any programme touching both halves needs them reported apart, because a paid placement and a named recommendation move for different reasons and stop for different reasons. AI Knows Us reads a question set you keep across ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews, and reports citation and naming as two separate columns, each engine on its own line. For a British buyer the plan that reads all five engines is 299 dollars a month, the price for the United States and the United Kingdom, read from aiknowsus.com on 22 October 2026. The page carries no pound figure, so a sterling number anywhere would be a conversion rather than a price.
For the British identification rules behind the first half of this page, see our page on advertising standards in the UK.