OpenAI now lets advertisers buy adjacency to a ChatGPT answer. For many founders, that is enough to trigger a test budget.
It should not be.
Read OpenAI's own documentation against buyer behavior data, and a quieter conclusion appears. The slot may be real, but it is not structurally positioned to reach the people most likely to decide a B2B software purchase.
The offer
OpenAI announced a self-serve Ads Manager on May 5, 2026, with CPC bidding, pixel-based measurement, and a gradual rollout for advertisers. In its announcement, OpenAI says Ads Manager is for “companies of all sizes, from SMBs and startups to global brands.” Source
That framing matters. It invites a budget decision, not just curiosity. If you are a startup with limited paid spend, the right question is not whether the product exists. It is whether the inventory can reach the right user at the right moment.
The help center line
OpenAI’s help center says ads may appear for users on the Free and Go plans, and that Plus, Pro, Business, Enterprise, and Edu accounts do not have ads. OpenAI Help Center
That is not a rumor. It is product policy.
It also changes the buying case. If your best prospects are using paid ChatGPT plans, then the ad slot cannot reach them. If they are not, the case becomes much easier to make.
Who researches
G2’s March 2026 survey of 1,076 B2B decision-makers helps define the audience. G2 reports that 51 percent now start software research with an AI chatbot more often than Google, up from 29 percent in April 2025. Seventy-one percent rely on AI chatbots for software research. Nearly two thirds spend six or more hours a week with chatbots at work. More than 40 percent are daily power users. Forty-one percent regularly use Deep Research tools for software evaluations. G2 survey summary
Two caveats matter. First, G2’s sample is global, not US-only. Second, the survey does not report which ChatGPT tier respondents use.
That second point is important. We are reasoning from usage intensity and employer provision to tier. This is an inference, not a measurement.
The inference
Here is the working argument.
If someone spends six or more hours a week on chatbots at work, uses Deep Research for vendor evaluation, and is a B2B software decision-maker, that user is likely to be on a paid plan provided by an employer. OpenAI’s own rules say paid plans do not see ads.
If that inference holds, the paid slot is blind to the exact cohort that decides many B2B software deals.
That is the core issue, not whether the placement is technically available.
Now the falsification test. If a large share of heavy professional users are actually on Free or Go, the argument weakens. G2 does not publish tier-level data. If someone has it, they should publish it.
What gets sold
OpenAI is also clear about what the slot does not buy. Its advertising principles say ChatGPT’s answers stay independent, and ads are clearly labeled as sponsored and visually separated from the response. OpenAI advertising principles
That is the right posture for a product platform. It also means the purchase is adjacency, not endorsement.
That distinction matters because G2 found that 85 percent of buyers think more highly of a vendor when a chatbot mentions them in a recommendation. Mention and adjacency are not the same product. Only one of them is for sale.
What buyers trust
G2’s data points to something more durable than ad placement. Forty-five percent of buyers say citations from software review sites are the most confidence-inspiring signal in an AI-generated response. Sixty-four percent encounter inaccurate recommendations often or very often. When a chatbot conflicts with a trusted brand, 24 percent go to peer reviews next. G2 survey summary
The practical reading is straightforward. Buyers are not just reading AI answers. They are checking the sources behind them.
That makes review coverage, comparison pages, and retrievable product language more valuable than a slot that sits outside the answer.
Why unknown wins
Two more figures matter. G2 says 33 percent of buyers purchased from a vendor they were not familiar with, and 69 percent chose a different vendor than they had initially planned, based on chatbot guidance. G2 survey summary
This is the uncomfortable part for incumbents, and the useful part for startups.
Being unknown is no longer fatal. A chatbot can surface a vendor the buyer had not planned to consider. That is a better argument for earning a mention than renting a slot beside an answer.
When buying fits
None of this means the ad product is useless.
There are cases where it makes sense:
- Consumer-adjacent products where the buyer is likely on Free or Go.
- Low-ticket software with self-serve motion and short evaluation cycles.
- Categories where you need a short-term floor while building citations and third-party proof.
- Launch tests where the main goal is learning, not efficient pipeline.
That is a narrower use case than many founders will want to hear. It is also the honest one.
Where money goes
If the buyer is not on an ad-supported plan, the same dollars should go somewhere else.
Start with the assets buyers actually trust:
- Earn reviews on the sites that get cited.
- Publish comparison pages that answer the question buyers are already asking.
- Make product pages retrievable by crawlers and useful to human evaluators.
- Write the objections into the page, not around them.
Those are not glamorous tasks. They are closer to revenue infrastructure than campaign theater, which is why they matter.
For a deeper look at the practical side of that work, see Better’s related coverage on why customer language now shapes AI answers, and why retrievability still determines whether answer engines can quote you at all.
Check before spend
If you are still considering a test, verify three things first:
- Whether your account is eligible to advertise in the US.
- Whether the current Ads Manager beta is actually open to your region and setup.
- Whether your most important buyers are likely using Free, Go, or a paid ChatGPT tier at work.
OpenAI’s Ads Manager beta overview says some capabilities are still limited and points advertisers to a separate availability page. Ads Manager Beta Overview
That is the right sequence. Eligibility first, buyer fit second, spend last.
People also ask
How to advertise on ChatGPT?
OpenAI has introduced a self-serve Ads Manager with CPC bidding and measurement tools, but the more useful question is whether the inventory reaches your intended buyer. For many B2B founders, the answer may be no. OpenAI announcement
Do ChatGPT ads show to paid users?
No, according to OpenAI’s help center, ads may appear for Free and Go users, while Plus, Pro, Business, Enterprise, and Edu users do not have ads. OpenAI Help Center
Are ChatGPT ads worth it?
They can be, but only in narrower cases than the headline suggests. If your buyer is likely on a paid plan, the slot cannot reach them. If your buyer is on an ad-supported plan, a test may be reasonable.
What is better than ChatGPT ads?
For B2B software, the higher-leverage spend is usually on trust assets, review generation, comparison content, and crawlable product pages. Those are the inputs that shape what AI systems cite and what buyers trust.
Takeaway
OpenAI now sells adjacency to the answer. That is real inventory, but it is not the same as reaching the buyer who decides the deal.
By OpenAI’s own tier rules, ads are not shown to Plus, Pro, Business, Enterprise, or Edu users. G2’s survey suggests the heaviest software researchers are the people most likely to use AI chatbots at work, run Deep Research, and sit inside employer-paid environments.
We cannot prove that all of them are on paid ChatGPT tiers. We can say the inference is strong enough to change the budget question.
Before you buy the slot, read the help page, test your buyer mix, and ask whether you are spending to be seen by the right person, or just spending to be near the answer.