Choosing a ChatGPT Ads Agency: What to Ask Before You Sign

Choosing a ChatGPT Ads agency is harder than choosing an agency for Google Ads, and the reason is simple. Nobody has years of results on the platform, so the usual shortcut of asking for case studies and sector benchmarks tells you very little. The better test is how an agency plans to handle policy, tracking and measurement on a channel that is still in beta, which is the standard Priority Pixels works to in its ChatGPT Ads services for B2B organisations.

The questions below work for any agency on your shortlist, including us. If you’re still getting to grips with the format, the guide to how ChatGPT Ads work covers the basics, and this piece assumes you’ve already decided the channel is worth testing.

Why a New Channel Raises the Bar for Agencies

A new advertising channel rewards agencies that are strong on fundamentals and honest about what they don’t yet know. ChatGPT Ads reached UK users in June 2026, according to Digiday’s report on the UK launch, so UK accounts have a few months of history at most.

OpenAI describes the product as in beta and in the early stages of scaling, and its advertiser FAQ states that no cross-advertiser performance benchmarks exist yet. An agency quoting typical click costs or conversion rates for your sector is working from something other than platform data, and it’s fair to ask where those numbers came from.

The platform has changed shape quickly during 2026 too. Trade coverage such as The Keyword’s pricing explainer tracks the move away from early minimum commitments and CPM only buying towards CPC bidding, so advice written in the spring may already be out of date.

What does carry over is paid media discipline. Conversion tracking, landing page quality, clean account structure and honest reporting decide whether a search or social account produces pipeline, and they decide the same thing here. Good ChatGPT Ads management is built on those habits rather than on platform experience nobody can have yet.

Eight Questions to Put to Any Agency

The most useful questions test how an agency will protect your account and your data before the first ad runs, not how confident it sounds about results. Ask them in a first meeting and request the answers in writing, so you can compare agencies on the same terms.

1. How will you check our category against OpenAI’s current ad policies, and which policy version will you record?
2. How will you set up the OpenAI Pixel and Conversions API, and how will they respect our cookie consent choices?
3. How will you confirm OpenAI’s crawlers can reach our landing pages before any ads are submitted?
4. What single success measure will we agree before any budget is spent?
5. How will you report results, and how often will we see them?
6. Who will own the advertiser account, the conversion data and the record of every setting?
7. How will ChatGPT Ads work alongside our Google Ads, Microsoft Ads and LinkedIn Ads activity?
8. How will you separate modelled conversions from conversions our website observed?

Vague answers to the first three are the most telling, since those checks decide whether an account can run at all. A strong agency will answer every question in concrete terms and point you to OpenAI’s own documentation where the platform sets the rules.

Policy, Tracking and Crawler Access Come First

A competent agency checks whether your category can advertise before it discusses budget. OpenAI’s ad policies prohibit most health and financial services advertising outside the US, with exceptions such as financial software, and bar legal services outside the US entirely. The policies are versioned and updated regularly, so an eligibility check is only as good as the date it was carried out.

Ask to see the policy version the agency checked against. Priority Pixels records it as part of every eligibility review and tells clients plainly when the platform isn’t open to them yet, which is the position it currently takes for UK healthcare, financial and legal advertisers.

Tracking is the next test. ChatGPT Ads measures conversions through the OpenAI Pixel, the Conversions API or a combination of the two, and OpenAI’s conversion measurement guidance tells advertisers to send conversion data only once users have been informed and every consent required by law has been obtained.

In the UK that points straight to PECR. The ICO’s guidance on how the storage and access rules apply to online advertising says consent covers any storage and access technology used for advertising, including measuring campaign effectiveness. The same consent questions shape PPC tracking without third party cookies on every other platform you run.

Landing pages are the third check. OpenAI requires its OAI-AdsBot crawler to reach every ad landing page for review, and its crawler guidance for advertisers lists firewall rules, CAPTCHAs, rate limits and login gated pages among the reasons pages fail validation. A good agency tests your pages against the crawler before submitting ads and gives your IT team the published OAI-AdsBot user agent and IP ranges to allow.

Reporting, Ownership and Your Other Paid Channels

Reporting is where a new channel most often gets oversold, so agree the success measure before any money is spent. For most B2B organisations a qualified lead, a demo request or a booked call recorded in your CRM is a far better measure than clicks or a platform conversion count.

Ask how the agency separates modelled conversions from observed ones. OpenAI’s conversion reporting can include modelled figures where available, which is reasonable for a platform but misleading if a report presents them as leads your website recorded. Priority Pixels shows observed and modelled conversions separately, alongside what your own analytics and CRM recorded, and clients can follow progress between monthly reports on the reporting dashboard.

Account ownership deserves a direct answer. OpenAI’s advertising terms require an agency to hold a written agreement with each advertiser it manages and leave the agency liable for those advertisers, which is a sound reason to insist the account sits under your own business details with the agency added as a user. If the relationship ends, you keep the account, the conversion history and a record of every setting.

Integration with your existing channels is the last part of this conversation. OpenAI’s ad group guidance is clear that context hints are not keywords, yet your search term data is still the richest record of how your buyers describe their problems. An agency that runs your Microsoft Ads and LinkedIn Ads accounts can judge the new channel against them rather than in isolation, and many of the points in our guide to choosing a Google Ads agency apply here unchanged.

Warning Signs That Should End the Conversation

Four warning signs suggest an agency is selling certainty the platform can’t support. OpenAI does name a small number of large holding group agencies as partners in its announcement of new buying options, and that is a matter of public record. Partner status isn’t needed to run ChatGPT Ads well, and the advertising terms bar customers from making public statements about their relationship with OpenAI without its written approval.

Treat any claim of special access, approval or endorsement with caution and ask for the source. The same scepticism applies to promises and price lists that the platform’s own documentation doesn’t support.

None of these signs on its own proves an agency is acting in bad faith, but each one should prompt a follow up question. An agency that answers with evidence, links to OpenAI’s own pages and admits what it doesn’t know is usually the safer choice.

What a Sensible First ChatGPT Ads Test Looks Like

A sensible first test has a fixed read period, one success measure, tracking proven before launch and a clear decision at the end. Priority Pixels runs ChatGPT Ads tests over a three month read period, with the conversion pixel and Conversions API in place and a test conversion proven before the first ad is submitted.

Budget should be set for the test rather than the platform minimum. OpenAI’s campaign setup guidance lists a minimum daily budget of £15 for UK accounts, but that figure describes what the system will accept, not what a B2B test needs to produce a readable number of leads.

Daily checks during the read period catch delivery, pacing and landing page problems early, and a monthly audit judges the account against its own baseline. At the end of the test you should get a straight recommendation with the numbers behind it, including a recommendation to stop if the channel isn’t producing qualified leads.

If you’re comparing agencies now, put the eight questions to each one in writing and read the answers side by side. For more on where the channel fits, the guides to ChatGPT Ads for B2B marketing and advertising on ChatGPT are useful next reads.