Two weeks ago, OpenAI opened ChatGPT ads to 26 more countries in a single move, the biggest jump since the test began. The scale question is settled now. The performance question isn't, and a study of nearly 10 billion sessions just gave the clearest answer yet.
What happened on October 21
OpenAI's own count puts the total at 89 countries, up from 63 the week before. Fourteen of the 26 new markets are in Latin America, including Argentina, Chile, Colombia and Peru. The rest reach into Africa, Eastern Europe and South and Southeast Asia, among them Nigeria, South Africa, Serbia, Ukraine, Cambodia, Bangladesh and Pakistan.
More than 100,000 advertisers have now run at least one campaign since the test launched on February 9. Ads still show only on the free tier and the $8-a-month Go tier. Plus, Pro, Business and Enterprise stay ad-free, which matters more than it sounds for anyone selling to a company account.
The scale is real, and so is the money behind it
This isn't a side project anymore. OpenAI's ad business reached a $1 billion annualized revenue run rate by August, up from $100 million in March, seven months earlier. The push lines up with a widely expected 2027 IPO, and a diversified consumer ad business is exactly the kind of story an IPO roadshow needs.
Buying in is simple on paper. Open OpenAI's self-serve Ads Manager, confirm your business is based in one of the 89 countries, and write your first creative. OpenAI hasn't published a minimum spend or benchmark costs for any market, so budget sizing is still a guess. L'Oreal's chief digital and marketing officer said testing early lets the company "pioneer helpful experiences and journeys, from discovery, to evaluation, to agentic commerce," which describes intent more than results.
200+ SaaS teams already track their AI citations.
They know exactly when ChatGPT mentions their brand, and when it stops. Do you?
What the first independent study found
A working paper by Maximilian Kaiser at Vienna University of Economics and Business and Christian Schulze at Frankfurt School of Finance and Management, published in October, tracked nearly 10 billion sessions across 1,898 online shops over 60 weeks. It's the first outside look at how ChatGPT ads actually convert, and it's still a working paper under review, not yet peer-reviewed.
In US traffic, a paid ChatGPT visit brought in $0.89 in revenue on average, against $2.45 for a paid Google Search visit. Conversion told a similar story. ChatGPT paid traffic converted at 0.37%, against 1.17% for Google Search. Oddly, ChatGPT's own organic traffic converted better than its paid traffic, at 1.31%.
The study followed 87 advertisers running both channels at once. Eighty-four earned less revenue per ChatGPT visit than per Google visit, and 73 recorded zero ChatGPT transactions despite nearly 19,000 paid sessions between them. The researchers calculated that a ChatGPT click would need to cost about $0.26 to match Google's return. One advertiser in the data paid $1.72 a click instead.
One number should temper all the others here. A single advertiser accounted for 66% of every ChatGPT purchase in the dataset.
Worth flagging too, the paper discloses that Kaiser advises Grips Intelligence, the firm that supplied the underlying traffic data. That's the kind of detail a careful reader checks for in any sponsored-adjacent dataset.
What advertisers are saying on Reddit
Reddit has been running its own messier version of this study for months. A thread collecting six months of real accounts shows both ends of the range.
ChatGPT ads- What I learned in the first 6 months
I've managed $50K+ in ChatGPT Ads budgets.Here are 6 observations from my experience so far CPCs are surprisingly familiar. In our campaigns, CPCs have been comparable to Google Search Ads for similar keywords and geographies. I expected a...
One agency spent about €1,000 each on three clients over two months, at roughly €2 a click, and found only 10 to 20% of visits were real engaged people rather than noise. They moved the budget to LinkedIn instead.
A different agency in the same thread works only with law firms and reported the opposite experience. One client spent about $1,000 and landed four personal injury cases. Another signed four cases on just $115. The agency was careful to say six weeks isn't enough data to call either result a pattern.
The thread's harshest critique came from someone who surveyed 160 people and found zero who had ever clicked a ChatGPT ad. That lines up with another commenter's claim that close to half of ChatGPT ad traffic looks like bots. Quoted costs per click across the thread ranged from about $3 to $13 for similar campaigns, with no agreement on why the spread is so wide.
One agency running enterprise B2B accounts alongside a regional ISP's consumer campaigns called the channel "incredibly hard to scale." They said it underperformed every other channel they run.
How often does ChatGPT mention your brand?
Most founders have no idea. The answer might surprise you.
Nobody has proven the ad model yet
The skepticism isn't limited to ChatGPT advertisers. One longtime SEO and PPC commentator summed up the mood on X days before the study went public, joking that Google and OpenAI should swap "I'm with stupid" T-shirts, since neither company has shown an ad model works when the entire point of an AI answer is to finish the task and leave.
We're like 3 years into this, and we still don't appreciate the level of crazy we're witnessing. Google goes all-in on AI Mode to chase ChatGPT (because, like Rodger Daltrey, they "won't get foooooled aaaagaaaaaiinnnn" - as in when Bing beat them to the punch). Now, OpenAI https://t.co/dvPPbrF2z2
Mordy Oberstein@MordyObersteinNov 3, 2026
That's a joke with real numbers behind it now. Google's own AI Mode ads just expanded to 10 more countries with no way for advertisers to opt out, and the click-through data Google has published so far is just as thin as OpenAI's.
Who should actually test this now
None of this means skip it. It means test the way the data suggests, not the way you'd test a channel you already trust.
- Cap the budget. Treat it as a small, fixed experiment rather than money moved off Google or LinkedIn until you have your own numbers.
- Pick the success metric before you spend. Cost per qualified meeting against your existing benchmark beats judging by raw clicks.
- Build one landing page with its own tracking link so you can trace which leads actually came from ChatGPT.
- Use manual bids for the first couple of weeks instead of automated bidding, and expect some days to return no ad at all. That's a normal outcome on inventory this new, not a broken account.
- Give it real weeks before judging. Pipeline data beats two days of click data every time.
B2B teams selling to Plus or Enterprise buyers should weigh all of this against one fact that doesn't change with better creative. Those buyers see zero ChatGPT ads on any budget, because the tiers they pay for stay ad-free.
The organic path still costs nothing to test
Every dollar spent testing whether ChatGPT ads convert is a dollar not spent on the channel that costs nothing regardless of budget. That channel is the organic answer, the brand ChatGPT already names without being paid to.
Our earlier look at ChatGPT's ad rollout found the same tier split holds here. The buyers actually signing SaaS contracts see no ads at all, so organic citations are the only lever that reaches them.
OpenAI also turned on automatic creative edits for every ChatGPT advertiser by default on October 30, which our coverage of that change walks through if you do decide to run a test.
RankControl's AI visibility tracking checks your citations across all seven engines, ChatGPT included, every week, at zero marginal cost no matter how your ad test goes. The real question was never just this one platform's economics. It's whether your citations are moving while you wait for the ad side to prove itself.
You can run the test plan above by hand and track your own citation rate in a spreadsheet next to it. Or RankControl's agents can handle the citation side every week while you run the ad experiment yourself.
15 hours a week manually. Or 15 minutes with RankControl.
Track citations, monitor competitors, and fix content gaps across every AI search engine. Automatically.
