What ChatGPT Ads on Shopify means for UK ecommerce brands

UK Shopify merchants can start buying ads in ChatGPT without leaving Shopify, after OpenAI began opening its ChatGPT Ads app to merchants outside the US on 23 September 2026. The UK already has self-serve access, so a brand can put its catalogue in front of ChatGPT users without an agency or a new product feed.

What OpenAI changed this month

The Shopify app arrived in a larger announcement on 16 September. OpenAI named Shopify its first ecommerce partner and HubSpot its first CRM partner, and said products are “already integrated through Shopify Catalog”, so a merchant can start running ads and tracking performance straight away. US merchants got the app that day, with other markets following from 23 September.

The same announcement added AI tools on the advertiser side. Advertisers can now create, update and analyse campaigns by prompting ChatGPT through an Ads Manager plugin, and Ads Manager itself suggests copy and imagery based on the landing page and the campaign objective. OpenAI also added an optional setting that rewrites an advertiser’s headlines and descriptions to fit each conversation and translates them into the user’s language. Sponsored Agents, which let a shopper who clicks an ad carry on the conversation with an agent the brand sponsors, are in a test with select US advertisers.

A week later, on 23 September, OpenAI took ChatGPT Ads into seven more Asian markets and put the total at more than 60 countries. It said the business reached $1 billion in annualised revenue in less than 200 days and that tens of thousands of advertisers now run ads in ChatGPT. Its availability page lists the United Kingdom for self-service. Digiday reported in August that UK advertising started in June on an agency-only basis, so British merchants have gone from needing an agency to buying from inside Shopify in about three months.

Who you can reach, and who you can’t

OpenAI shows ads only to people on ChatGPT’s Free and Go plans, and keeps Plus, Pro and Enterprise ad-free. For a consumer brand that covers most shoppers. For anyone selling to senior buyers who use a paid ChatGPT plan at work, ads won’t reach them at all, and the only way into those conversations is still being the source ChatGPT cites. We covered how far citations have drifted from Google rankings in The citation moved off the page.

OpenAI also says ads never influence ChatGPT’s answers. Buying ads won’t get your products recommended in the answer itself, so you have to run paid and organic as separate efforts, even though both appear on the same screen.

What changes for UK ecommerce teams

The ads draw on whatever sits in your Shopify catalogue, so your product titles, descriptions, images and prices do the job ad creative used to do, and a thin or inconsistent catalogue now costs you in a paid channel as well as in organic AI answers.

Our view at Manual Focus is that a UK Shopify brand with clear product-level demand should run a small, controlled test this autumn, paid for from its experimental budget rather than cut from paid search or social that already performs. We’d leave the automatic text customisation off at first, because it rewrites your headlines in your name and you should see what it produces before you let it run. Paid channels are turning into contests between AI systems on both sides, which we wrote about in Paid channels in the AI era.

How we’d run the first test before Black Friday

1. Tidy the catalogue first. Fix product titles, descriptions and images in the categories you plan to advertise, because the ads draw on them.
2. Start with one category. Choose one where shoppers ask detailed questions before they buy, such as sizing, compatibility or use cases, since those are the conversations ChatGPT handles.
3. Test the automatic copy separately. Run a second ad group with text customisation switched on, and compare what it writes and what it converts against your own copy.
4. Measure against paid search. Tag every click, and compare cost per order and returning-customer rate with your Google Shopping campaigns for the same products. The attribution tear-down playbook shows how to separate genuine lift from double counting.
5. Keep working on organic citations. Nobody on a paid ChatGPT plan sees the ads, which leaves citations as your only way into those conversations.

What’s still unclear

OpenAI said the Shopify app reaches other markets “starting September 23”, so availability may roll out over several days, and merchants should check the App Store listing from their own account. There’s no public UK performance data yet, no date for Sponsored Agents outside the US, and OpenAI says new formats, optimisation tools and measurement are still to come.

If you’d like a second pair of eyes on a test plan, get in touch.

I run ecommerce marketing for [brand], [one line on what you sell], on
Shopify, selling mainly in the UK.

We're about to test ChatGPT Ads through the Shopify app. The ads draw on our
Shopify catalogue, so product titles, descriptions, images and prices do the
job ad creative used to do. We're testing one category: [category], where
shoppers ask detailed questions before buying, such as [sizing /
compatibility / use cases].

Attached is a product export for that category with titles, descriptions,
image alt text and prices.

For each product, check whether:
- the title says plainly what the product is
- the description answers the questions shoppers ask in this category
- titles and descriptions are consistent across similar products
- image alt text describes the product

Give me a table with the columns: Product | Problem | Suggested new title |
Suggested new description (under 80 words).

Don't add specifications, sizes or claims that aren't in the export; write
"needs input from [team]" where a fact is missing. List any products or
columns you couldn't read.

We're testing OpenAI's optional setting that rewrites our ChatGPT Ads
headlines and descriptions to fit each conversation and translates them into
the user's language. I want to check what it writes in our name before we
let it run more widely.

Attached are:
- the rewritten headlines and descriptions from our test ad group
- our original ad copy
- our brand guidelines: [file name]

Go through the rewritten copy line by line and flag anything that:
1. changes or adds a product claim compared with our original copy
2. drifts from the tone in our brand guidelines
3. looks mistranslated or unnatural in the target language

Give me a table with the columns: Original line | Rewritten line | Issue
(claim / tone / translation / none) | Why | Suggested fix.

Finish with a one-paragraph summary of whether the rewrites are safe to keep
running. If you can't judge a translation with confidence, say which language
and leave that row for a native speaker to check.

Set this up as a routine that runs every Monday morning.

We're testing ChatGPT Ads for [brand] in the UK against our Google Shopping
campaigns for the same products in [category]. Every click is tagged.

Each week I'll attach:
- a ChatGPT Ads results export for the past week
- a Google Shopping results export for the same products and dates
- our orders export with a returning-customer flag

For each channel, work out cost per order and returning-customer rate, by
product and in total.

Give me a table with the columns: Product | ChatGPT Ads cost per order |
Google Shopping cost per order | ChatGPT Ads returning-customer rate | Google
Shopping returning-customer rate | Change since last week.

Under the table, list what moved most since the previous week in three lines
or fewer. If an order can't be matched to a channel, or a product appears in
one export and not the other, list it separately rather than guessing.