Latest · Ad platforms

ChatGPT Ads Still Has No Idea What Good Costs

September 8, 2026

Six months into ChatGPT Ads and advertisers still can't tell you what a good result looks like. CPCs are landing anywhere from single digits to $22, and there's no auction insight, no industry benchmark, and no audience data solid enough to explain why.

For a lease-up team, that spread isn't a rounding error. That's the difference between a channel worth scaling and a $22 lead for a studio that rents for $1,400. And right now nobody, not the platform, not the agencies, not the case studies you've seen floating around LinkedIn, can tell you which one you're going to get.

Here's the move: if you're testing ChatGPT Ads, treat the budget like an R&D line, not a media line. Small, fixed, and walled off from your real lead targets. Don't compare its cost-per-lead to Meta or Google, because you have nothing to compare it to yet. That's not a knock on the platform, it's just where it is. Auction transparency and category benchmarks take time to build, and right now you're flying with no instrument panel.

The failure mode is obvious: someone gets a great CPC in month one, calls it a channel, and shifts real budget into it before there's any pattern to trust. One good week is a fluke wearing a strategy costume.

Keep testing. Just don't confuse a test budget for a proven one, and don't let anyone above you do it either.

Source: Search Engine Journal, September 8, 2026