A firsthand look at OpenAI's new advertising platform
When OpenAI dropped its $50,000 minimum ad spend on their new beta advertising platform, we moved quickly to get a live campaign running. We saw it as a huge opportunity to get real experience with a new advertising channel before it became crowded and competitive. Not because we expected ChatGPT to replace Google or Meta ads overnight, but because the way people search, compare options, and make decisions is changing rapidly and advertising is starting to follow them into those new behaviors. Plus, new advertising platforms are fun!
We really wanted to roll up our sleeves and get real campaign data. So we launched a client campaign (with their support) and built the supporting measurement around it, spent real budget, and watched what happened for about a month. With anything new, we really had no expectations and wanted to see how the platform could perform with minimal adjustments and really let it take the wheel.
Now, after 30 days, we have a much better feel for how OpenAI Ads actually work, where the platform still has some growing to do, and what advertisers should know before jumping in. Some of our findings were surprising and the platform will only get better.
NOT Just Another Paid Search Channel
The biggest adjustment was not learning where the buttons were (that was actually very user-friendly). It was changing how we thought about the campaign itself and what defines it.
Google Ads has trained marketers to think in terms of keywords, match types, search volume, and bidding signals. Advertising inside ChatGPT starts from a completely different place and mindset. The surrounding conversation and the intent behind it matter much more. That changes how you think about audience, messaging, and when an ad is actually useful.
That may sound simple, but it has important strategic implications. You cannot simply transfer a Google Ads campaign to a new interface. Instead, you must build around how people use ChatGPT to research problems, explore options, and ask follow-up questions. Don’t think, “What are they searching?” instead think, “What are they asking?”
The First 30 Days
We had no previous data to compare success and LinkedIn Gurus on these metrics are all over the place with hardly any opinions lining up. Our first campaign was intentionally a test, but it generated enough activity to give us something meaningful to evaluate success.
- 17,074 impressions
- 362 clicks
- 2.12% click-through rate
- $2.69 average CPC
- 2 legitimate leads traced back to the campaign
For a new channel, the CPC immediately stood out. More importantly, we were able to trace actual lead activity beyond the ad click. That matters because inexpensive traffic is not useful if it disappears the moment someone reaches the site.
Two leads are not enough data to declare ChatGPT the next major acquisition channel, and we would not pretend otherwise. But it was enough to show that the traffic had real commercial intent and was worth continuing to test.
Fast Decisions
One of the more interesting things we saw was how quickly delivery concentrated around one message. The platform found a clear direction early and moved a large share of traffic toward it.
That was noticeably different from the learning periods we are used to seeing in Google and Meta. After one campaign, I would not call that better optimization. It could also mean the system is too aggressive about picking a winner. But it is exactly the kind of behavior you only start to understand after running real campaigns.
It also reinforced why early testing matters. The advantage right now is not having every answer. It is building enough experience to recognize what the platform is doing, question it when needed, and know what to test next.
Measurement matters even more on a new ad platform
We also learned quickly that evaluating performance only inside the OpenAI Ads dashboard would not be enough.
A new advertising channel needs the same discipline as an established one: clear conversion actions, clean attribution, and a way to follow traffic through to actual leads. We built that measurement foundation before relying on the campaign results, which let us separate useful activity from vanity metrics.
That work is now part of how we approach OpenAI Ads going forward. The platform itself will keep changing, but the need to understand what happens after the click does not.
What changed during the first month
One of the challenges with being early is that the product can change while you are actively using it. When we launched, the campaign options available to us were very limited. The only options for campaign goals were Clicks or Impressions. By the end of the first test, the campaign goal of Conversions was now out of beta and available as an option.
That is a good example of why we wanted to start now instead of waiting for the platform to feel finished. The version of OpenAI Ads advertisers use six months from now will almost certainly be different from the one we first tested. Having a working knowledge of the platform makes those changes much easier to evaluate and adopt.
OpenAI Ads Worth Testing?
For the right advertiser, yes. I would not recommend pulling budget out of a proven Google Ads program just to chase a new channel. But I do think OpenAI Ads deserve a place in the testing conversation, especially for companies whose buyers spend time researching services, comparing providers, and asking detailed questions before they ever fill out a form.
That is where ChatGPT is already part of the customer journey. Paid placements create a new opportunity to show up inside that research process.
The bigger takeaway from our first 30 days is not any single performance metric. It is that we now have real campaign experience. We have seen how the platform behaves, where the setup differs from traditional paid media, what needs to be measured outside the ad dashboard, and what we would do differently on the next campaign.
Want to test ChatGPT advertising for your business?
If you are curious whether OpenAI Ads make sense for your business, GRAYBOX can help you evaluate the opportunity and run a structured test without treating your budget like an experiment. We can handle the strategy, campaign setup, landing page experience, measurement, and ongoing optimization based on what we have already learned from running the platform firsthand.
Get in touch and we can talk through where ChatGPT advertising could fit into your paid media strategy.
There is still a lot to learn, and OpenAI Ads will continue to evolve. That is exactly why we wanted to be early.
