ChatGPT ads for ecommerce: what they cost, and when a $4 click pays back.

OpenAI's own help pages still say ChatGPT ads have no performance benchmarks across advertisers, industries or campaign types. Six months in. Ad platforms rarely admit that, and it means nobody pitching you this channel knows what good looks like yet, us included.

What we do know is the price of a click. For an ecommerce brand that's enough to run the math before spending a dollar, and the math is less forgiving than the launch coverage.

01 How do ChatGPT ads work?

ChatGPT ads are sponsored placements shown below ChatGPT's answers to adult users on the Free and Go plans. Each one is labeled Sponsored and sits apart from the answer. OpenAI says the ad systems do not influence what ChatGPT says, and that conversations stay private from advertisers. Plus and Pro subscribers see no ads at all.

Matching runs on what the conversation is about. Ads stay away from accounts OpenAI predicts belong to under-18s, and from sensitive topics such as health and politics. That rule matters more than it sounds if you sell supplements, or anything shoppers ask about in medical terms: expect thin inventory.

Stores can run product ads from a catalog feed, which entered the Ads Manager beta in June, next to regular link ads. You buy on cost per thousand impressions or, since May, cost per click. All of this sits under the answer. Getting named inside the answer is a separate job, covered in our AI search playbook.

02 The 2026 timeline, from test to self-serve

OpenAI announced an ads test in January, put the first ads live in February, opened a self-serve Ads Manager with no minimum spend in May, added product feed ads in June and reached 31 European markets in late August. Each step made it cheaper for a mid-size store to try.

The price of entry went from $50,000 to zero in one announcement. The price of a click didn't come down with it.

03 How much do ChatGPT ads cost?

OpenAI recommends a starting maximum bid of $3 to $5 per click. The first named tests, gathered by Search Engine Journal in September 2026, averaged $4.41 to $9.89 per click, and advertisers buying impressions reported CPMs above $60. With no official benchmark yet, read every figure below as one account's experience.

For scale, our Meta benchmarks roundup puts the 2026 all-industry Meta CPC near $1.72. A ChatGPT click at $4.41 has to be worth about two and a half Meta clicks. It might be. Someone who clicks out of a conversation about the exact product is further along than someone scrolling a feed.

04 The break-even math for a $4 click

A click pays for itself on the first order when order value × contribution margin × conversion rate is at least the cost per click. Turn it around and the smallest order that breaks even is CPC ÷ (margin × conversion rate). At $4 a click, a 40% contribution margin and a 4% conversion rate, that's a $250 order.

Why 4%? Adobe found that AI-referred visits to US retail sites converted 42% better than other traffic in March 2026, after converting 38% worse a year earlier. Contentsquare's 2026 benchmark puts ecommerce paid search sessions at about 2.8%, as we covered in our conversion rate benchmarks. Add 42% to 2.8% and you land near 4%. Adobe measured AI referrals in general, not ad clicks, so treat 4% as a hopeful starting guess.

The smallest first order that breaks even, at a 40% contribution margin:

If your average order is $70, no cell comes close. That doesn't rule ChatGPT out. It means the first order can't carry the click, so repeat orders have to, and you need your repeat rate and payback window before you bid. We laid out that method in how much to spend on ads, and our calculator gets you contribution per order in a few minutes.

Then there's attribution, which can flip the verdict on its own. Common Thread Collective reported $19,000 to $38,000 in attributed revenue on the same $9,620 of spend, depending on the attribution model. Assume a 40% contribution margin. The low reading loses about $2,000. The high one makes about $5,600. Only the model changed. When the platform's column is the only judge, the platform is marking its own homework, the trap we took apart in why your ROAS is lying to you.

05 Who should test ChatGPT ads now, and who should wait

ChatGPT ads suit stores with high order values and products people research by asking questions. They don't suit small one-off baskets, or brands whose Meta and Google accounts still lose money. This is the checklist we run before recommending a test.

Test now if:

• Your average order clears your cell in the table, or repeat purchases pay back a smaller first order inside your window.
• Shoppers research your product with questions before they buy: mattresses, strollers, cameras, espresso machines.
• Your product feed is clean and your best sellers have reviews worth reading.
• You already send server-side events to Meta or Google, so adding OpenAI's Conversions API is an afternoon of work.

Wait if:

• Your orders are under about $100 and customers rarely come back.
• Your main channels aren't profitable yet. A new channel won't fix a leak in the offer or the page, so fix the chain first.

06 The first test we would run

Set up measurement before the first dollar, bid at or below your break-even CPC, buy enough clicks to read a conversion rate, and judge the result on your own store data. Around 500 clicks, roughly $2,000 at $4, is the smallest test we'd trust.

1. Install OpenAI's pixel and send the same purchases through its Conversions API. OpenAI deduplicates on pixel ID, event name and event ID. Set your click window before launch. View-through conversions use a fixed one-day window and are reported apart from the main Conversions column; we wouldn't let them decide the verdict.
2. Start with product feed ads on your highest-value products with the best reviews. Bid at or below the break-even CPC from your cell of the table. If you can't win clicks at that price, that's your answer for this quarter.
3. Buy about 500 clicks over three to four weeks. At 4% conversion that's 20 orders: enough to tell a 4% channel from a 1% channel, not enough to fine-tune anything.
4. Judge on contribution. Compare new-customer orders and blended MER for the test weeks against the weeks before. If you sell in more than one market, holding out a country is cleaner; the method is in our incrementality testing playbook.
5. Write the rule down before you start. Scale only if the test makes money on contribution under the conservative attribution model, or pays back inside your window on repeat orders.

07 Where to start

Start with your average order value and contribution margin, and find your cell in the table. If a $4 click can pay there, run the 500-click test with measurement in place, and keep it separate from your Black Friday budget, which has its own math. If you want a second opinion on whether ChatGPT belongs in your mix at all, send us your numbers. And if your average order is $70, what bundle would get it to $250?

Sources: OpenAI, Testing ads in ChatGPT (January 16, 2026) and Our approach to advertising and expanding access; Axios on the February 9 launch and the May 5 self-serve Ads Manager; OpenAI and Quartz on the dropped $50,000 minimum and CPC bidding; OpenAI Ads developer documentation on the Conversions API, attribution windows and deduplication; Search Engine Land on product feed ads (June 2026) and the 31-market European expansion (August 2026); Adweek and eMarketer on Criteo's $10,000 minimum; Search Engine Journal, 6 months into ChatGPT ads (September 8, 2026), for the Common Thread Collective, Synter and Hostinger figures and OpenAI's $3 to $5 bid guidance; Adobe Digital Insights AI traffic report (April 2026); Contentsquare 2026 Digital Experience Benchmark.