Advertising & Media

Gaps in OpenAI's Measurement Hold Back Ad Budgets in ChatGPT

Seven months after the launch of ads in ChatGPT, marketers complain about attribution failures: the platform shows zero conversions despite real leads and overstates clicks fivefold.

1 min read

OpenAI’s measurement gaps are keeping ChatGPT ads budgets at test level
OpenAI’s measurement gaps are keeping ChatGPT ads budgets at test levelAI-generated

Seven months after ChatGPT began selling ads, marketers still don't understand how to measure the effectiveness of these placements. The lack of clear analytics has been the main stumbling block in nearly every discussion of OpenAI's progress as an ad seller.

The frustration has two causes. The first is the sense that OpenAI has built almost nothing beyond basic functionality. The second is advertisers' reluctance to entrust their data to the company.

Discrepancies in the numbers

At the product level, agencies running campaigns describe gaps that would look unusual for Google or Meta. UK agency Accuracast ran a lead generation campaign in the UK exclusively in ChatGPT and received completed forms from users. Yet two weeks later, the platform still showed zero conversions.

"The problem with Google and Meta is usually the opposite — the platforms claim more conversions than they actually delivered," noted Farhad Divecha, group CEO of Accuracast.

The same agency also found a discrepancy in clicks: ChatGPT reported around 100 click-throughs, while the agency's own tracking recorded 20. OpenAI's support service was unable to explain the reason for the discrepancy at the time.

Not everyone is struggling

The problem is far from universal. Executives at AdRoll, Jellyfish, and another independent US ad agency said they had not encountered discrepancies in click data.

For Web Guide Partner, the main difficulty is the speed of data delivery. Where conversions in ChatGPT are in fact registered, the agency finds they arrive with a delay of 24 to 36 hours. By comparison, Google delivers data within a few hours.

Another limitation is that reporting contains no demographic data or information about users' prompts.

Taken together, these limitations keep ad budgets in ChatGPT at a test level: marketers are unwilling to scale their investments until they get reliable attribution tools and confirmation that reported numbers are accurate.

Source: Digiday

Share this article:

More from Сергей Мельник

Related articles

« Previous articleNext article »