Does paying change what ChatGPT says about you?
Not according to OpenAI. Ads run on separate systems, and the answer is not for sale. So the answer still has to be earned, through clear, direct content on your own site and consistent facts across the web. That is answer engine optimisation, and it matters more now, because a weak answer above your ad undoes the ad. Our piece on answer engines covers the method.
Who sees the ads?
Only Free and Go users. Many enterprise buyers use ChatGPT through paid business plans, which carry no ads. For B2B brands, the earned answer reaches decision makers that ads cannot. For consumer brands, the reach among free users may be the bigger opening. Test before you commit budget.
How should you measure it?
OpenAI says new formats, optimisation tools and measurement are still to come. Until they mature, measure what you control: send ad traffic to dedicated pages, capture every lead in the CRM and judge the channel on pipeline and revenue, not clicks. HubSpot's AEO tool, in beta, tracks brand visibility in AI answers alongside.
What does this mean for budget planning?
Treat ChatGPT ads as a test line in next year's plan, not a reallocation. The format is new, measurement is still being built, and the audience is limited to free users. Fund the earned side first. Direct answers on your own site, consistent facts on the pages AI assistants read, and tracking of how you appear in AI answers all pay off whether or not you buy ads. Then set aside a small, fixed budget to test paid placement in one or two markets, with a clear pipeline target and a stop date. If the test produces qualified pipeline at a cost you would accept from search or social, scale it. If not, the earned work still stands, and you have learned cheaply.
What to do on Monday.
- Ask ChatGPT the ten questions your buyers ask about your category, and note whether you are named.
- Fix the gaps with direct answers on your own site before you buy any ads.
- Work out whether your buyers mostly use free or business plans.
- If you test ads, run a small, time-boxed pilot with its own landing page and CRM source.
- Report the test on pipeline created, not impressions.