Whether it's worth testing for you
Before anything is built, the same questions I ask when a client forwards me OpenAI's invite email. They decide more than any benchmark, and there isn't a reliable UK benchmark yet anyway.
I'm encouraging every client I already work with to at least consider it, especially now that it can target a local area and OpenAI is offering new advertisers a credit to try it. Considering it is where these questions come in.
If the answer is to wait, you'll hear that, and why. The detail behind each question is in ChatGPT Ads in the UK: worth testing yet?, and where it sits next to Google and Microsoft is in the three-platform comparison.
The ad buys the card below the answer, not a mention in the answer itself. That part is organic, and earning it is covered in how to get your business mentioned in ChatGPT's answers.
- Is your category allowed? Healthcare, financial services, legal services, gambling, alcohol, dating and political ads are excluded. If that's you, no budget changes the answer for now.
- Does the targeting fit how you sell? There are no keywords to bid on. You describe the conversations your ad belongs in and OpenAI decides where it shows. Targeting below country level has only just reached UK accounts, so I check what the location picker offers in yours before we plan anything.
- Can you afford the test? At this stage it should still be treated as a testing budget. Where ChatGPT Ads has high potential for a business, that testing budget can be chunkier.
- How would you feel about losing it? A first test can come back with nothing but what it taught you. That has to be survivable before it starts.
What I do
Built from nothing in your own account, measured before a pound is spent, and judged on what a customer costs rather than what a click costs.
- An Ads Manager account in your name and on your billing, with me invited as a user. You can remove me whenever you like
- The OpenAI pixel and Conversions API installed together and deduplicated before the first pound, so the test ends in results rather than screenshots
- OpenAI's crawlers, OAI-AdsBot and OAI-SearchBot, let through your robots.txt and firewall. OpenAI's rules require it, and an ad whose landing page they can't reach fails review
- A campaign built from scratch rather than from the draft OpenAI writes for you. Budget, objective, landing page and context hints are decisions, not defaults
- Context hints written as briefs: the situation your buyer is in, in several specific versions. Where you already run Google Ads, the search terms that convert are where they start
- A landing page that matches the conversation the ad appears in, not your homepage
- Location set as tightly as your account allows, confirmed in the picker rather than taken from the help centre
- AI-written copy variations left off unless you're happy for wording neither of us has read to run under your name
- Results in the same monthly report as Google and Microsoft, measured on blended cost per customer
- A keep-or-stop decision at 60–90 days, against a number we agree before the first ad runs
What running it has shown me so far
I'm running ChatGPT Ads on my own account through the £500 spend-match offer OpenAI began emailing UK advertisers in late August. I started on 6 September 2026, and the results will be published when the spend completes.
Two things turned up before any of it was spent. The campaign OpenAI drafted for me arrived with an ad it couldn't finish, next to a £50 daily budget, a clicks objective, my homepage as the landing page and no context hints at all. It left the creative undone and set every number that costs money, which is why I build campaigns from scratch.
The same day, the location picker in my UK account accepted postcode districts in Leeds and London, six days before OpenAI's help centre said UK advertisers could target below country level. For a local business, the account is the thing to check, not the documentation.
Both are written up in full, with what they do and don't prove, in Should you take ChatGPT's £500 ad credit?
What it costs
A flat monthly fee, whether ChatGPT Ads is the only platform I run for you or one of three.
Flat means it doesn't move when your ad spend goes up or down, so nothing about it rewards me for growing the test beyond what the numbers support. It does move with how much there is to manage. Adding ChatGPT Ads to Google or Microsoft costs more than running one platform, but it doesn't double: the more of your advertising I run, the less each platform costs you.
The ad spend is paid to OpenAI from your own account and never passes through me. The fee is quoted after the intro call; how it is set is on Services.
If OpenAI has emailed you a credit offer, it changes the arithmetic but not the questions above. Read your own email for the terms, because they differ between waves; ChatGPT Ads credit, explained sets out what the published terms do and don't commit to.
Common questions
Do I own the account?
Yes. It's your own Ads Manager account, in your name and on your billing, with me invited as a user. OpenAI gives every advertiser an account of its own, and you can remove my access whenever you like.
Can I hire you for ChatGPT Ads alone?
Yes. I run any one of Google, Microsoft and ChatGPT Ads on its own, or any combination of them. The ad spend goes to OpenAI from your own account.
Does adding it cost more?
Yes, because there is more to manage. But the fee doesn't double when a second platform is added, and it doesn't grow with your ad spend on any of them. The more of your advertising I run, the less each platform costs you.
How much should a first test spend?
It should still be treated as a testing budget at this stage. For some businesses ChatGPT Ads has high potential, and there that testing budget can be chunkier. We'd settle the figure for your business on the intro call.
How soon will I know if it works?
Judge it at 60–90 days, on what a customer costs across all your platforms together rather than on the price of a click. The number that decides it is agreed before the first ad runs.
Worth a test for your business?
Thirty minutes, free. Whether your category is allowed, whether the targeting reaches the people you sell to, and whether to test this quarter or wait.
