By Sara Khan · WeProms Digital · November 6, 2026
Every marketing voice a small British boutique owner hears in November 2026 — the agency decks, the LinkedIn screenshots, the Golden Mile WhatsApp groups where everyone swaps ad tips — repeats the same instruction: move £500 of next month’s ad budget into ChatGPT ads now, while the clicks are cheap, the auction is thin, and the big retailers have not yet arrived to crowd you out.
The instruction deserves a colder reading, because the only transaction-level comparison published so far points the other way. A working paper by Maximilian Kaiser and Christian Schulze — “Can Advertising Fund Free AI? Evidence from ChatGPT’s International Advertising Rollout,” October 2026 — measured paid ChatGPT visits generating $0.89 in revenue per session against $2.45 for paid Google Search visits, with conversion rates of 0.37% versus 1.17%. In the best evidence that exists, a paid ChatGPT click is worth roughly 36% of a Google click. The pattern repeats across every new ad surface: reach arrives years before evidence does, and the bill for the gap lands on whoever moved first. This essay is the spreadsheet a Leicester abaya boutique should run before a single pound moves.
Reach is real, and it is not the question
The bull case is not invented. According to Similarweb and Graphite data reported by Digiday, the UK has the highest proportion of ad-eligible ChatGPT users actually seeing ads, at 45.7% — ahead of the United States at 43.4% — and ChatGPT ads only switched on for UK users in early June 2026. For a Leicester boutique whose customers ask ChatGPT where to buy a black abaya or a jilbab for Eid, being nameable inside that answer is genuinely worth something.
So what does 45.7% prove? It proves delivery, not outcome. Fewer than half of eligible UK users see ads at all, and nothing in the figure says whether an ad produces a sale. A reach statistic and a revenue statistic are different instruments; conflating them is how early-mover advice gets written. The UK figure says the stage is lit. The Kaiser-Schulze data says the actors are being paid in pennies.
There is also a practicality the LinkedIn posts skip: buying these ads as a small UK advertiser is still not a self-serve checkout in the way Google Ads is. OpenAI’s rollout began with managed placements and partner integrations, and eligibility for independent boutiques has not been documented the way a Golden Mile owner would need it to be. Checking OpenAI’s current advertiser documentation before budgeting anything is not caution; it is the first gate.
Six thousand brands is a crowd, not a verdict
The fear-of-missing-out case leans on growth numbers, and the growth numbers are real. Adthena tracked 1,082 distinct brands advertising on ChatGPT in June 2026, 3,741 in July, and 6,333 in August across the US, Australia, the UK and the EU. New Street Research models commercial chats on the platform growing from 45.9 billion this year to nearly 380 billion by 2030. Numbers like these are presented as proof the train is leaving the station.
Read them again, slower. Six thousand entrants in a single summer is not evidence of returns; it is evidence of experimentation budgets. Digiday’s reporting on Sensor Tower’s US data shows shopping advertisers’ share of ChatGPT spend falling steadily since spring while financial services and software grow — the category that rushed in first is the category most willing to fund tests, and its share is shrinking, not sticky. What actually drives this is not proof but pressure; an owner on the Golden Mile hears “everyone is in” and reads it as “everyone is winning,” when no published figure says any small retailer has won anything.
Shaadi season teaches the same lesson every year. Everyone agrees you book the photographer fourteen months out — early is the whole strategy. But booking early only pays if the hall is right; the couple that locked the photographer into the wrong venue paid twice. Early money committed to the wrong channel behaves exactly the same way.
What $0.89 a session does to a £72 order
Run the math on a real basket. Worked example, illustrative — substitute your own costs: a Leicester boutique’s average abaya order is £72; product, packaging, and postage come to £41; contribution per order — price minus product and fulfillment costs — is £31. At the study’s 0.37% conversion rate, one order demands roughly 271 paid clicks. New Street Research models ChatGPT CPC — cost per click, the price an advertiser pays each time someone clicks their ad, per Google’s own definition — at around $5, a figure its analysts themselves caution may be too high. Take the generous end anyway: at roughly £3.90 a click, those 271 clicks cost about £1,055 to win £31. Even at a hypothetical £1 click, 271 clicks still means £271 to earn £31.
Put Google through the identical math before anyone calls this unfair. At the study’s 1.17%, an order needs about 86 clicks; at that same £1 click, £86 to win £31 — still nearly three times the contribution, which is precisely why mature Search accounts survive on branded demand and remarketing rather than cold clicks alone. Cold paid traffic is expensive everywhere. On ChatGPT it is currently about three times more expensive per order than on the channel that already demands discipline.
One honesty note that cuts both ways: the Kaiser-Schulze figures describe US advertisers on ecommerce websites, and no UK transaction study exists yet to confirm or contradict them. A Leicester boutique is not a US dataset. But the direction of the finding — paid chatbot sessions converting at roughly a third of search’s rate — does not hinge on geography, and prudence says you size the risk on the evidence you have, not the evidence you hope for.
The tracking gap costs more than the clicks
Even a boutique willing to burn tuition money on a test faces a second problem: the test will not produce a trustworthy answer. Farhad Divecha, group CEO of Accuracast, told Digiday that the latest partner integrations, Magnite and PubMatic among them, have not resolved the underlying measurement problems:
Some are teething problems because it’s a new system that has to be set up on advertiser systems, but some problems are down to attribution being messed up.
For a one-till shop on Belgrave Road, that sentence is the whole risk. A boutique without server-side tracking cannot verify whether a ChatGPT click ever became an order; the dashboard will claim credit for sales the till never saw, or miss sales it caused, and either error poisons the next budget decision. Search Engine Journal made the adjacent point in its November piece on AI visibility versus search rankings: agencies are still celebrating metrics that do not answer the buyer’s question. Citation counts and mention shares flatter the report; they do not ring the till. An advertiser who cannot measure revenue per session is not testing a channel — they are donating to it.
When the answer flips
None of this argues ChatGPT ads will never work. It argues the entry conditions are not met for a store with a £72 average order and no analyst. Three gates, in prose. First, evidence: a published UK transaction study, or your own Shopify order data, showing chatbot sessions converting within striking distance of search — the gap today is a factor of three, and it needs to close before the unit economics clear. Second, measurement you trust: conversions traceable to actual orders at your order values, verified before scale, not after. Third, contribution headroom: your contribution per order must cover the worst-case acquisition cost from the published data, not the best-case from a pitch deck — on today’s numbers that means the average order needs to be in the hundreds of pounds, not £72.
Until the gates open, the budget has honest work elsewhere: a clean Google Shopping feed, and the post-purchase email flow that monetizes buyers you already paid for — the same retention discipline UK mithai shops learned when stamp cards failed them. Calendar the review honestly: Ramadan 1448 is expected around February 8, 2027, subject to moon sighting, and Eid gifting peaks in the two weeks before the expected Eid date of March 10, 2027, per Islamic Relief’s calendar. Re-run this math in January 2027 before the season locks in — gift demand is exactly when impulse-friendly surfaces earn their fairest test, and when new evidence, if it exists, will have had a quarter to surface. If you want the cost-side view of running a test despite the math, read what a first ChatGPT ads test actually costs a halal food brand — then come back to the conversion rate before signing anything.
The principle is older than every platform that has ever promised cheap attention: spend where each session’s revenue is both measured and positive, and enter new channels only when your unit economics survive their worst published numbers rather than their best promises. Cheap clicks are not a strategy; they are a price. The boutiques that win the AI-search era will be the ones that treated it as a spreadsheet question first and a story second.
Read next: Why does ChatGPT get my halal grocery’s details wrong? and Why do my Ramadan customers disappear after Eid?
WeProms Digital has managed paid media for halal grocers, modest-fashion stores, and takeaways across the UK, Gulf, and North America — the budget math above is the same discipline we apply before recommending any channel, new or old. If you are weighing ChatGPT ads against Google Shopping, our ChatGPT ads management service builds entry tests around contribution-per-order math, not hype; and if the honest answer today is “not yet,” our PPC audit team will tell you that and show you where the budget earns more. Based on our experience, the owners who win are the ones who asked for the math first. Reach us at hello@weproms.com or WhatsApp +92 300 0133399, or through weproms.com/contact-us.
Sources & References
- Digiday — In graphic detail: Six thousand brands later, ChatGPT ads still have a conversion problem — November 5, 2026
- Kaiser & Schulze — Can Advertising Fund Free AI? Evidence from ChatGPT’s International Advertising Rollout (study page) — October 2026
- PPC Land — ChatGPT ad visits earn shops $0.89 against $2.45 for Google Search ads — November 2026
- Search Engine Journal — AI Visibility vs. Search Rankings: Is Your Agency Measuring the Wrong Thing for 2027? — November 5, 2026
- Google Ads Help — About cost per click — accessed November 2026
- Islamic Relief UK — Islamic calendar and Ramadan dates — accessed November 2026
- timeanddate — Eid al-Fitr dates — accessed November 2026
Additional reading from industry feeds:
