OpenAI opened its ad platform to advertisers in 2026. It's small, it's unfinished, but we think it will be an important channel in the future. That's why we're testing it now with clients who can afford to learn something before everyone else does.
Before anything launches we agree on what would make this worth continuing, in your numbers: a cost per qualified lead, a number of them, and a date.
The ChatGPT ad unit is small: a title of about 16 characters, roughly 32 characters of copy, your logo, an image, and a link. There's no room to explain yourself at length, so the writing has to be efficient.
OpenAI's pixel and Conversions API wired up properly, tagging so ChatGPT shows up as its own source in your analytics and your CRM, and the attribution window set deliberately rather than left on the default.
There are no keywords here. Placement runs on the context of the conversation, with geographic exclusions and bid weighting where they help.
Customer list matching exists but needs about 25,000 matched people to run, so for most local businesses it isn't on the table yet. We'll tell you if you're one of them.
Nobody has years of performance data on this channel, including us. What we do have is a way to find out cheaply, with a controlled budget and with an exit strategy.
Most of what we learn here will be useful to you later, when ChatGPT Ads are bigger and the people who waited to jump in are starting from zero. Opening an ad channel early can have huge benefits (think of Google in 2002).
You don't have to hire us on retainer to get value. If you want to run this yourself, you can just buy the expertise by the hour.
For teams who want to test this in-house: account setup, tracking, what to write for the format, and an honest read on whether your numbers mean anything yet. $200/hr, no minimum commitment.
Book time →For most businesses, yes. It's early for anyone who needs this budget to produce leads. It's worth doing if you're already covered on the channels that work and you'd rather learn a channel before it gets expensive.
Yes. The hourly consulting above exists for exactly that. No retainer required, no pressure to upgrade.
Clicks currently run in the range of $3 to $5, and OpenAI removed the spend minimum it required during the managed pilot, so there's no platform floor to clear. We ask for $3,500 a month in ad spend as a minimum, because below that the result won't have enough volume behind it to tell you anything.
No, it's not. Ads appear below the answer, marked as sponsored, and they're kept separate from the model's response on purpose. You cannot buy your way into being recommended. Getting mentioned in the answer itself is a different problem, and it isn't one you can solve with ads.
There's a lot of things you can't measure. There's no user-level data, no demographic breakdown, and no reliable way to credit someone who saw your ad, kept researching in the chat, and turned up later as a direct visit. Clicks and conversions we can track properly. Influence we largely can't, which is another reason to treat this as a test rather than a core part of your ad strategy.
Yes. That's the whole point of agreeing on the win condition before we start. If the number isn't there at 90 days, we'll recommend moving the budget back to what's working, and we'd rather do that than keep a line item alive because it sounds impressive.
Book a free 30-minute call. We'll tell you honestly whether your business is a candidate yet, and it's fine if the answer is no.
Book a Discovery Call