If you run an Amazon brand, the news that ChatGPT started serving ads in the UK on June 6 probably hit your feed as "OpenAI is becoming Google." That framing will cost you. The thing to watch is not the ad unit. It's what an ad unit confirms: a few hundred million people now start product research inside a chat window instead of a search bar, and OpenAI just put a price on appearing in that conversation. That changes where demand forms — upstream of your Amazon listing, before the shopper ever types your category into the search box.
I'm not telling you to go buy ChatGPT ads this week. Most of you shouldn't. I'm telling you the discovery layer moved again, the same way it moved with Rufus and Alexa for Shopping, and the operators who treat this as "another channel to evaluate" are answering the wrong question.
What happened
On June 6, 2026, OpenAI expanded its ChatGPT advertising pilot to the United Kingdom — the first market outside North America and Oceania — with Japan, South Korea, Brazil, and Mexico queued up next. Ads show only to free and Go-tier users, in labeled placements, with context-based targeting. This sits on top of the self-serve Ads Manager OpenAI opened to any U.S. business on May 5 with no minimum spend. The pilot reportedly hit $100M annualized within six weeks and OpenAI is projecting $2.5B in ad revenue for 2026.
Two sentences of news. The rest is what it does to you.
Why most brand owners will read this wrong
The dumb take: "A new paid channel just opened. I should test ChatGPT ads before my competitors do."
No. For a brand doing $50K–$500K/month on Amazon, ChatGPT ads in June 2026 are an early, thin, English-first pilot with primitive targeting and no Amazon attribution. If you chase it now you'll spend money you can't measure against a funnel you don't control, while your actual conversion problem — the listing — sits untouched. First-mover advantage on an unproven ad surface is mostly a story people tell after the fact.
The real signal: OpenAI doesn't monetize a behavior that isn't already massive. You put ads where the eyeballs already are. The launch is the receipt, not the event. The event happened months ago, quietly, when your customers started asking "what's the best [your category] for [their problem]?" inside a chatbot and acting on the answer. That shifts a chunk of top-of-funnel discovery off Google and off the Amazon search bar — and into a system that returns one synthesized recommendation, not ten blue links or a grid of thumbnails.
A grid gives you a fighting chance with a better hero image. A synthesized answer gives you a fighting chance only if the model knows your product exists and has a reason to name it. That's a different game, and it's the one that matters.
What actually changes for someone running $200K/mo on Amazon
Nothing changes in your ad account this month. Here's what changes in how you should think:
Branded search becomes a leading indicator, not just a defense line. When a shopper asks ChatGPT "best non-toxic cookware for a small kitchen" and the answer mentions your brand, the shopper doesn't buy inside ChatGPT — they go search your brand name on Amazon. So AI discovery shows up in your data as rising branded search volume with no matching ad spend or social push. If you're not watching Search Query Performance for your branded terms as a demand signal, you're flying blind to the one place AI discovery is already measurable for you. I've started treating an unexplained branded-search lift as evidence the assistants are recommending a client — and it's checkable today, for free, in Brand Analytics.
Your "machine-legible" attributes are now doing double duty. The same structured detail that helps Rufus shortlist you on Amazon helps an external model describe you accurately. Material, use-case, dimensions, who-it's-for, what-it's-not — written as plain factual claims in your bullets, A+, and Brand Story — is the raw material these systems quote. Pretty lifestyle copy that implies benefits gives the model nothing to repeat. This isn't a new project; it's the project you should already be doing, now with a second payoff.
The cost of being un-namable goes up. On a search grid, a generic listing still gets impressions — it's in the results whether or not it's distinctive. In an AI answer, the un-namable product simply isn't mentioned. There's no "page 2" of a ChatGPT answer. The penalty for blending in moved from lower CTR to zero inclusion. That's a real margin event over 18 months, and it compounds quietly.
What it does NOT change: your CVR, your ACOS, your hero image's job. The shopper still lands on your PDP and still decides in about three seconds. ChatGPT can send you a more qualified visitor, but it can't close them. If your listing converts at 9% today, a better discovery channel just sends more traffic to a 9% listing. The leak is still the leak.
What I'd do this week if I were you
- Pull 90 days of branded search volume from Brand Analytics / Search Query Performance. Set a baseline now, before the assistants ramp. You want to be able to say "branded search is up 14% and I didn't spend a dollar to cause it" three months from now. That number is your AI-discovery dashboard.
- Audit your top 5 ASINs for factual, quotable attributes. Open each listing and ask: if a model had only this text, could it accurately say who this is for and why it wins? Rewrite anything that's vibe instead of fact. This is the same attribute-fill work I've been pushing for Rufus — do it once, get paid in two channels.
- Ask ChatGPT, Gemini, and Rufus your category's three highest-intent questions. "Best [category] for [use case]." See who gets named. If it's never you, that's your real competitive gap — and it's free to check.
- Leave your ad budget alone. Do not move money into ChatGPT ads. If you have genuine off-Amazon DTC infrastructure and a measurement stack, you can run a tiny $500 test purely to learn the interface — framed as tuition, not a channel. Everyone else: skip it.
- Write down a trigger. Decide in advance what would make ChatGPT ads worth real budget — e.g., "shoppable product placements with Amazon attribution" or "available in my market with conversion tracking." When the trigger hits, move. Until then, watch.
What I'd ignore
Ignore the "$2.5B in ad revenue" headline — it tells you about OpenAI's business, not yours. Ignore the breathless "ChatGPT will kill Google Shopping" takes; the relevant question for you isn't who wins the ad-platform war, it's whether your product gets named in answers, and that's true regardless of which engine wins. Ignore the agencies that will pitch "ChatGPT Ads management" by August — there is almost nothing to manage yet for a marketplace brand, and paying a retainer to manage a $500 learning budget is how you fund someone else's experiment.
And ignore the urge to do something because the news felt big. The right move when a discovery layer shifts is rarely a new ad campaign. It's making sure the thing being discovered — your product, described in language a machine can repeat — actually holds up when the spotlight finds it. That work was already on your list. The news just raised the interest rate on skipping it.