You’re used to ChatGPT ads being just text and a logo.

They can be pictures now. OpenAI just released a bunch of updates, including a visual ad format in the US, shown while Free and Go users wait for an image to generate.

The ads are labeled and kept separate from the image being created. What's still unclear is how they'll be targeted, and whether someone making an image is in a buying mood.

And you can finally get some real assessment. OpenAI's measurement expansion connects to the tools you already run, and the partner list is long:

• Conversion data via Hightouch, Tealium and LiveRamp.
• Attribution via AppsFlyer, Triple Whale, Northbeam, Adjust and more.
• Geo incrementality tests with Haus, Measured and WorkMagic.

WeightWatchers is the first proof point. According to DV Rockerbox, its CPA came in 15.3% below its blended paid-search benchmark.

Brand safety also gets a new control. Qualifying advertisers can now add Negative Phrases to keep ads away from topics their own policies rule out. Nice.

And organic products are getting badges. ChatGPT is testing superlative labels, with a green "best all-rounder" tag and grey ones like "best for beginners" or "best cushioning."

Plus, the ad slot is testing a friendlier intro: a new heading that reads "This might also interest you." It frames sponsored placements as suggestions.

There's no performance data for these ads. Yet. But better measurement means we should have some info on that soon as well.