OpenAI moves to make ChatGPT ads more measurable — and more visual

Since OpenAI started selling ads in ChatGPT in February, advertisers have kept coming back to the same complaints: they can’t measure the ads properly, they don’t have enough formats to work with and there isn’t enough inventory to absorb what they want to spend. OpenAI’s latest round of updates takes aim at two of them.

First there’s the ad format.

For the past seven months, advertisers have been working with a relatively simple ad format which includes the advertiser name, logo, headline, description, landing page and image. But later this month, OpenAI is planning to start testing image ads in the U.S.

The initial test will run during image generation, with ads clearly labeled and kept separate from the images ChatGPT creates. The idea is to give advertisers another way to show people what a product or service could actually look like, rather than solely relying on a text ad.

Next, is OpenAI’s measurement update. 

As Digiday recently reported, measurement is one of key pieces OpenAI needs to unlock bigger ad budgets. Without the ability to connect ChatGPT campaigns to outcomes, advertisers are likely to stick to contained test budgets. 

To do so, OpenAI is now expanding its in-house tools and partner integrations with the likes of Kochava, along with 10 other partners, so advertisers can use tools they already rely on to assess web and app outcomes, such as results beyond the last click. Geo-based incrementality, for now, is still in early experimentation with partners including Haus, Measured and WorkMagic.

And then there’s brand suitability.

OpenAI is also working with DoubleVerify and Integral Ad Science on pilots to evaluate brand suitability in controlled testing environments. But in the spirit of privacy, the platform said those pilots won’t give DoubleVerify or Integral Ad Science access to real user conversations.

Taken together, the updates reflect how early OpenAI’s ad business remains: the company is still building out the formats, measurement tools and safeguards that advertisers typically expect from a more mature ad platform.

While these updates address some of advertisers’ biggest concerns, one issue is harder to solve: inventory.

“There are about 2.5 billion prompts globally happening each day on ChatGPT, whereas there’s about 15 billion searches on Google so the volume is inherently smaller,” said Zeno Group’s evp paid social, Shamsul Chowdhury. “Compare that to the fact that there’s only one ad placement on ChatGPT whereas Google has multiple, it all leads to a disconnect between supply and demand.”

Still, these latest updates to the ChatGPT ads offering comes at the start of Advertising Week New York, where OpenAI execs will be using the event to sound out marketers and get their feedback on the product so far.

More in Marketing

‘The future is going to be in real life”: IRL events are becoming the new proof point for creator marketing

Plenty of creators are making the same bet, and until recently most of them were the biggest names in the business.

‘We’re all going to live with uncertainty forever’: Advertising Week New York opens amid industry rebuild

Why this year’s Advertising Week New York feels different (yes, we know we say that every year).

Marketing’s upper middle embraces AI creative production

Scaled AI creative production is becoming ordinary among CMOS outside the billion dollar club.