ChatGPT ad entry minimums dropped from $250K to $0 in four months. Here's what's independently verified, what's vendor-stated, and how to test this channel without overcommitting.
When OpenAI's ChatGPT ad pilot launched on February 9, 2026, the reported entry commitment was $200,000 to $250,000 at a $60 CPM. Four months later, StackAdapt offers entry at $0 minimum, Criteo at $10,000, and OpenAI direct at $0. Over 2,000 brands are now advertising on ChatGPT through Criteo alone, with rollout expanding from the US to Australia, Canada, New Zealand, the UK, Japan, and South Korea.
That minimum-spend compression is a genuinely useful signal. It tells you how badly every ad network wants a foothold in this channel while it's still early. It does not, by itself, tell you whether advertising there is actually a good idea for your business yet.
This is the distinction that matters most, and it's the one most coverage of ChatGPT ads skips entirely. Reach is real and well documented. Whether the performance numbers being used to sell you on the channel are equally solid is a separate question.
Criteo reports that users referred from LLM platforms like ChatGPT convert at roughly 1.5 times the rate of other referral channels, based on a sample of 500 Criteo retailers in the US. That's a real, if narrow, dataset, and it's Criteo's own client base being used to sell Criteo's own product. Criteo's Prompt Smart Ads capability separately claims early tests generating approximately 4x higher results, though the specific metric and methodology behind that figure isn't independently published.
OpenAI itself projects $2.5 billion in ad revenue for 2026. Independent analysts haven't confirmed that figure. The closest outside corroboration is a Barclays research note from April 2026 projecting $2.4 billion, which is close enough to lend some credibility to OpenAI's number, but it's still a projection, not a measured result.
Here's the part of this story that rarely makes it into the pitch decks: broad consumer survey data shows most US adults say ads inside AI chatbots would reduce their trust in the platform. eMarketer's own forecast expects AI-chatbot ads to represent only about 3 percent of total AI-related ad spend in 2026. The reach is real. The audience's actual appetite for being advertised to inside a conversational AI tool is, at best, unproven and at worst actively negative.
This tension, real reach paired with skeptical consumer sentiment, is exactly the kind of nuance that gets lost when a headline conversion statistic from a vendor's own client base gets repeated without its source attached.
Given the falling entry costs, testing ChatGPT ads is now accessible to businesses that couldn't have justified the original $250,000 commitment. That accessibility is the genuine opportunity here, not a promise of results. Enter through whichever partner offers the lowest-friction test for your situation, since minimums now range from $0 to $10,000 depending on the platform.
Demand CPC or performance-based pricing where it's available rather than committing to a flat CPM buy, since a fixed CPM structure puts all the risk of an unproven channel on you. Treat the first campaign explicitly as a structured test with a defined budget ceiling and a clear success threshold tied to your own break-even ROAS, not to a vendor's self-reported conversion lift.
Businesses with an existing, well-optimized presence on Google and Meta, looking to test genuinely incremental reach with a small, bounded budget, are the most sensible candidates for an early test. Businesses currently struggling with core channel performance are better served fixing that foundation first. A new, unproven channel rarely fixes a problem that exists in a business's fundamentals, and the cost of testing here, while lower than four months ago, is still real spend against an unverified return.
Not yet, for most businesses. The falling entry costs make testing accessible, but eMarketer's own forecast of roughly 3 percent of AI ad spend suggests even the industry's own analysts see this as a small, early channel rather than a core one in 2026.
Treat vendor-reported statistics as a reasonable starting hypothesis, not a guarantee. A 1.5x conversion lift from a 500-retailer sample within one company's own client base is real data, but it's not independent, peer-reviewed measurement, and your own results may differ meaningfully.
Enter through a platform offering CPC or performance-based pricing rather than a flat CPM commitment, set a firm budget ceiling for the test, and measure results against your own break-even ROAS rather than a vendor's self-reported benchmark.
This connects directly to the same evaluation discipline covered in our 90-day paid media audit and our breakdown of the ROAS formula every new channel should be measured against. If you're weighing whether an emerging channel like this fits your actual budget and goals, that's exactly the kind of question a paid media partnership is built to help answer honestly.
How We Work
