OpenAI is beginning to price some ChatGPT ads on a cost-per-click basis instead of a cost-per-impression model, which is how it launched the offering, according to The Information sources.
The source also said that OpenAI plans to introduce ads aimed at getting people to take a specific action like making a purchase or downloading an app, though the company hasn't put a timeline on when that could happen. Wait, its original ads weren't designed to get people to take an action? Does OpenAI know what “ads” are?
Since launching ChatGPT ads in early February, OpenAI has only provided early advertisers with general data such as impressions, clicks, and spend, which has made it difficult to tell if the ads are working. Some experts believe that this performance ambiguity is holding many advertisers back from advertising with OpenAI.
To support the shift toward performance-based advertising, Digiday reports that OpenAI is building a conversion tracking pixel that fires a signal back to the ad platform when a user completes an action on an advertiser's website after seeing an ad. The pixel is already live for select advertisers in the pilot and supports event types including lead created, order created, page viewed, subscription created, and trial started.
The immediate challenge with OpenAI launching a tracking pixel is that the technology is already outdated, as JavaScript pixels are often blocked by privacy-first browsers and ad blockers, which is why every major advertising platform has had to build alternate tracking solutions like server-side tracking. However, at the moment, there's been no information released on if / how OpenAI plans to tackle server-side tracking.
Honestly, it feels like OpenAI has continually been putting the cart before the horse with its advertising endeavors, collecting checks from partners before it's had the technology or process in place to best serve them. In general, they seem very lackadaisical with their partnerships — making promises that they don't deliver on, backpedaling on product releases, and ultimately wasting their partners' time and money as if they've all just had their own $110B funding rounds. The “move fast and break things” model worked in the late 90s and early 2000s, but the industry has matured. Companies expect more from their tech partners, and as of now, OpenAI hasn't delivered.