Enterprise

OpenAI is testing a persistent ChatGPT Ads identifier that could finally separate paid clicks from organic AI referrals

A cross-session tracking signal in testing would let small-business advertisers prove which ChatGPT spend converts — closing the attribution gap that has kept cautious marketers on the sidelines.

Photo: Unsplash / Emiliano Vittoriosi — ChatGPT app icon displayed on a smartphone screen

OpenAI is testing a persistent, third-party-style identifier inside ChatGPT Ads, MarTech reports, and the implication for small-business advertisers is larger than the mechanic itself: it would finally split paid clicks from organic AI referrals inside GA4, where both currently arrive under the same chatgpt.com source.

That single ambiguity has kept cautious marketers at 5-to-30-person shops on the sidelines of a platform that OpenAI has now pushed into more than 60 countries, including a Sept. 23 rollout to Indonesia, Malaysia, the Philippines, Singapore, Thailand, Vietnam, and Taiwan. You can’t optimize what you can’t attribute. You especially can’t defend a media line to a founder when the referral column doesn’t distinguish the ad you paid for from the mention you didn’t.

The rest of the September stack is already in place. Conversion-optimized impression billing (oCPM) went generally available on Sept. 16, alongside a HubSpot integration, and the Shopify app expanded outside the US on Sept. 23. Sponsored Agents entered a US-only test the same day. Ads serve only to Free and Go tier users, per Fit Small Business; Plus, Pro, Business, and Enterprise remain ad-free. OpenAI recommends a $3 to $5 starting max bid for CPC.

What’s missing is measurement rigor. Fit Small Business notes OpenAI hasn’t published performance benchmarks across advertisers, industries, or campaign types, meaning there’s no authoritative average CPC, CTR, or ROAS to judge against. Qwestyon’s September rundown adds that ChatGPT Ads now supports view-through conversions, adjustable 7-, 14-, or 30-day click windows, a 0- or 1-day view window, and an Event Quality Score grading pixel setup from 1 to 10. Those changes tend to inflate reported totals, and OpenAI’s own help center acknowledges figures may include modeled conversions.

So the practical move for teams already evaluating ChatGPT Ads as a discovery channel is to install the ChatGPT Pixel and Conversions API in tandem now, set the view window to zero for cleaner cross-channel comparison, and audit Event Quality Score before the identifier ships. Clean pixels on the day of rollout is worth more than a quarter of retroactive reconciliation.

The identifier isn’t confirmed. But the platforms that eventually beat measurement skepticism, Google in the mid-2000s, Meta in the early 2010s, did it by shipping attribution first and inventory second. OpenAI appears to have read the same playbook.

Sources