Practical examples: what a test looks like in three UK sectors
These are illustrative scenarios built to show the method. They are not client results.
Example 1: An online homewares retailer
The retailer already runs Google Shopping. First, confirm the product feed meets OpenAI's requirement for shopping placements. Then test context hints such as "someone furnishing a first rented flat on a tight budget". Judge the test on new-customer share, not return on ad spend alone. For a sense of what advertisers report, Triple Whale says 93% of Portland Leather's visitors from ChatGPT Ads were new. That is a US, partner-reported figure, so treat it as a prompt, not a forecast.
Example 2: A B2B software firm
A CRM provider has a six-week sales cycle. The context hint describes the conversation: "an operations manager comparing CRM tools for a team of ten". Success is measured in qualified pipeline inside the CRM, because form fills alone will flatter a channel that attracts early-stage researchers.
Example 3: A regional accountancy firm
The firm wants consultations booked, not clicks. Hints focus on specific worries, such as "a landlord unsure how to report rental income", and the landing page names the city. Call tracking and a source field on the booking form show which enquiries began in ChatGPT.
How to measure ChatGPT ads and avoid common mistakes
Measure ChatGPT ads against your search results, not against hope. Ads Manager reports impressions, clicks, spend, click-through rate, average CPC, average CPM and conversions, according to AI Tools Review.
Metric | What it tells you | How to read it |
CPC and CPM | The price of attention | There is no UK benchmark yet. US reports range from under $3 up to $10 or $13 a click, and Hallam saw CPMs as high as £60 in some sectors |
Click-through rate | Ad relevance | A low rate usually points to vague hints or a weak headline |
Landing page conversion rate | Page match | Compare with search traffic landing on the same page |
CPA against blended search CPA | Pass or fail | Your main decision metric |
New-customer share | Reach beyond existing demand | A high share suggests the channel finds people search misses |
Incrementality | True lift | Use a geo or holdout test once volume allows |
Five mistakes to avoid
Judging on clicks alone.
Copying a keyword structure. Context hints are not exact-match keywords, and OpenAI says they do not guarantee delivery.
Pausing search to fund the test.
Forgetting UK ad rules. The ASA's CAP Code still applies to AI ads.
Expecting an ad to earn a recommendation. OpenAI says ads do not influence answers, so organic visibility still needs good content.
ChatGPT, Claude and the wider AI ads landscape
ChatGPT is not the only assistant your clients will mention. Anthropic has said that Claude will remain ad-free: no sponsored links beside conversations, and no advertiser influence on responses. So there is no Claude ad inventory to buy, and anyone pitching it should be asked to show their working.
Claude still matters to paid media teams. Customers use assistants such as Claude and ChatGPT to research purchases, so clear, well-structured website content helps your client be understood wherever people ask. Our post on how AI tools are changing SEO agencies covers that side.
AI ads are a category, not one platform. Google is also pushing AI-led formats such as AI Max, and other assistants may follow OpenAI, so revisit your test plan each quarter.
Whether you are hiring or briefing, a capable partner sounds sceptical in the right places. Put these five questions to any digital marketing specialist or paid media agency before they touch your budget:
How will you design the test, and what is the pass mark?
How will you measure beyond the ChatGPT dashboard?
How will you protect existing search and social performance?
Which parts of the platform are still beta, and which are proven?
How do you handle UK advertising rules and brand safety?
Our guide to the five-minute test for a PPC or paid media agency is a useful companion.
How GSS Tech Solutions can help
GSS Tech Solutions is a UK digital agency with offices in London and Leeds. Our paid advertising service covers Google Ads, social ads, remarketing, PPC audits and SEA consulting, with a process that runs from audience analysis to A/B testing and reporting. Our data analytics and data visualisation work helps when a new channel needs joined-up reporting. If you are unsure whether a ChatGPT test suits your business, a PPC audit is a sensible place to start.
Frequently Asked Questions (FAQs)
Are ChatGPT ads live in the UK?
Yes. OpenAI's update confirms ChatGPT Ads has launched in the UK, and Ads Manager Beta lists the UK as available.
How much do ChatGPT ads cost?
You buy on CPC or CPM. OpenAI suggests starting CPC bids at $3 to $5, and Harrison Carloss reports a US average CPC of around $3.42. There is no reliable UK benchmark yet, so set your own pass mark.
Do ads change what ChatGPT says?
OpenAI says they do not, and that advertisers cannot read conversations. That promise has not yet been independently audited at UK scale, according to AI Tools Review.
Does Claude show ads?
No. Anthropic has said Claude will remain ad-free.
Will ChatGPT ads replace Google Ads?
No. Current UK agency advice is to test ChatGPT alongside search, because Google captures active demand while ChatGPT reaches people earlier in their research.
Final Words
ChatGPT ads are live, the audience is large and the UK evidence is still thin. That combination calls for a test, not a leap. Choose clients whose customers research before they buy, ring-fence a modest budget, set the pass mark before launch and judge results against your search CPA.
A confident paid media agency does not chase every new channel. It tests the right ones, for the right clients, in the right order. If you would like a second opinion on whether ChatGPT ads belong in your plan, talk to GSS Tech Solutions or call 0203 599 8122 for a free quote.