ChatGPT is starting to look less like a chatbot and more like a storefront.

Building the store: OpenAI just rolled out a batch of ChatGPT Ads updates, mostly for e-commerce:

• Bulk product-feed campaigns from a spreadsheet upload. Ads Manager builds the ad templates for you.
• Product review status that shows why OpenAI rejected individual products.
• Conversions beyond your campaign goal, pulled through the Insights API.
• New markets across Southeast Asia and Taiwan.

OpenAI says product-feed ads perform best on the platform so far, so catalog-heavy brands have the most reason to test.

Now stocking the shelves: ChatGPT shopping results also picked up virtual try-on and Favorites globally. Shoppers upload a selfie, hit Try On, and save what they like.

That moves ChatGPT onto Pinterest and Google turf for style discovery. OpenAI already shelved Instant Checkout after it underperformed, though.

The receipt is the problem: Agencies say measurement gaps keep budgets at test level.

One saw zero platform conversions two weeks into a lead-gen run, and about 100 reported clicks against 20 it tracked.

Others see conversions land 24 to 36 hours late, with no demographic or prompt data. Many legal teams also refuse to sign OpenAI's tracking terms.

So set up your own measurement before you test, and judge results by your own data.

Until OpenAI's numbers catch up, keep ChatGPT spend in the test column.