ChatGPT's new visual ads and measurement stack: what performance marketers actually get

What it is

A new image-based ad unit inside ChatGPT, plus a measurement layer bolted onto the existing ChatGPT Ads auction. OpenAI will begin testing visual ads during image generation in ChatGPT later this month in the U.S. with an initial group of advertisers. On the measurement side, OpenAI's pixel and Conversions API solutions measure actions such as purchases, leads, installs, and sign-ups, and integrations with Hightouch, Tealium, and LiveRamp let advertisers send conversion data from their existing systems to ChatGPT Ads.

Attribution partners sit on top. OpenAI is integrating with Hightouch, Tealium, and LiveRamp for conversion data, while attribution partners now include AppsFlyer, Adjust, Branch, Triple Whale, Kochava, and several others. Brand suitability is in pilot with DoubleVerify and IAS as the announcement details.

The next-work-session test

You run paid acquisition for a DTC brand. Your Monday session is "where do I put the next $10k test budget?" Until now, ChatGPT Ads was a reach bet with thin reporting. This update means you can wire your existing Triple Whale or AppsFlyer setup into a ChatGPT campaign and get back something close to the attribution view you already use for Meta and TikTok — same partner, same dashboard. That changes the test from a leap of faith to a line item you can defend in the next pacing review.

If you're not already in the pilot cohort, the next-session change is smaller: you're reading the roadmap and deciding whether to request access.

Pricing

The announcement itself does not publish rates. From third-party coverage of the broader ChatGPT Ads auction: OpenAI activated cost-per-click buying in April 2026. Advertisers pay only when a user clicks, and OpenAI recommends a starting max bid of $3–$5 per click. For reach buys, OpenAI sets a default max bid of $60 CPM. That $60 is a bidding ceiling in a second-price auction, not the price you pay. Minimum daily budgets are published at $25 USD and £15 GBP.

The visual format and new measurement tools themselves: no separate fee disclosed. Treat the pilot as free to join but gated by advertiser access.

What we'd actually use it for

A controlled geo-lift test on a visual-heavy category — apparel, home goods, travel — where the image format earns its placement next to a user-generated image. Pair it with the Conversions API so you're not relying on last-click. The honest pitch is incrementality measurement on a channel most of your competitors haven't figured out yet, not scale.

Limits

• US only, invite-only cohort at launch, image-generation surface only. OpenAI is also working with DoubleVerify and Integral Ad Science on brand suitability, which is particularly important for ChatGPT because ads appear around conversations rather than a traditional webpage. Translation: brand-safety tooling is still being built in public.
• No ads to minors and no ads in Temporary Chats, per public documentation of the broader program — so your addressable inventory is narrower than ChatGPT's headline user count.
• Attribution partners are listed but each integration has its own setup lift. Budget a sprint, not an afternoon.
• Creative production is on you. A visual unit next to AI-generated imagery sets a quality bar that stock product shots will lose.
• No published benchmarks yet for CTR, CVR, or CPA on the new format. Anything a rep tells you in the next 60 days is directional at best.

Try it if

• You already run paid social and have a Triple Whale, AppsFlyer, Adjust, Branch, or Kochava setup you can extend.
• You sell a visually demonstrable product and have fresh creative in rotation.
• You have budget earmarked for incrementality testing, not just scale.
• You're US-based and willing to request pilot access.

Skip it if

• You run lead-gen B2B with long sales cycles — the measurement window won't flatter you yet.
• You don't have a conversions API or server-side tagging in place. The measurement upgrade is only as good as what you feed it.
• Your creative team is already behind. The format rewards production quality.
• You need published CPA benchmarks before you'll test a channel.