ChatGPT ads get third-party attribution as OpenAI expands ad measurement

OpenAI published a post on Monday titled "Building advertising for the way people use AI". Its own summary says the company is introducing "a new visual ad format in ChatGPT" and expanding "measurement tools, attribution partnerships, and brand suitability for advertisers." The same day, ad-verification company DoubleVerify (DV) announced a related piece of that measurement push: ChatGPT Ads can now be tracked inside its attribution product, and DV is building a brand-suitability pilot with OpenAI.

Taken together, the two announcements show where ChatGPT advertising is heading. The question advertisers keep asking about a new channel is not whether it gets clicks, but whether it produces customers they would not have won elsewhere, and whether their ads appear next to content they can live with. Monday's news addresses both, with the caveat that the only performance number so far comes from a single advertiser, measured by a vendor that sells the tool.

What DV is launching

According to DV's release, advertisers can now add ChatGPT Ads as a touchpoint inside DV Rockerbox, the company's attribution and outcomes-measurement product. The point is to evaluate ChatGPT alongside the other channels a brand already buys, rather than reading the channel's own dashboard in isolation.

• ChatGPT Ads can be measured as one touchpoint in a multi-channel customer journey
• Attribution covers acquisition, conversions and return on ad spend (ROAS)
• The capability sits inside DV's Media AdVantage Platform (DV MAP), next to its verification and optimization tools

DV cites one early result. Weight Watchers added ChatGPT Ads as a member-acquisition channel and used DV Rockerbox to measure it. DV says ChatGPT's attributed cost per acquisition was 15.3% below Weight Watchers' established benchmark during the measured period. The release does not give the length of the period, the spend, or the attribution model used, so this is one data point, not evidence that the channel beats others in general. Weight Watchers itself calls it "an encouraging early signal" while it evaluates how to scale.

The brand-suitability pilot

The second half of the announcement is still in development. DV and OpenAI are building a brand risk and suitability evaluation pilot meant to let independent partners assess and help inform OpenAI's suitability policies and guardrails "in a controlled testing environment, without accessing private user conversations."

Conversational environments are fundamentally different because context is dynamic. What a user asks, how the AI responds and how the conversation evolves can all shape the environment in which an ad appears.

That framing matters. On the open web, verification firms classify a page and decide whether an ad belongs there. In a chat, there is no fixed page: the context is generated turn by turn and is private. The pilot's stated design, where the verifier tests OpenAI's policies in a sandbox instead of reading real conversations, is a trade-off. It protects users, but it means advertisers get assurance about the rules, not an audit of each placement. DV gives no launch date for the pilot.

What is still unclear

OpenAI's own post was not accessible to us from our research environment, so this article does not describe the visual format beyond the company's one-line summary in its news feed. We have not independently confirmed details reported elsewhere, such as where in ChatGPT the format appears, when testing starts, or which other partners are involved. We will cover those once they can be checked against a primary source.

For marketers, the practical takeaway is narrow but real: if you already use DV Rockerbox, ChatGPT Ads can now be compared against the rest of your media mix in the same tool. For everyone else, Monday's announcements are a sign that OpenAI is building the standard ad-industry plumbing (third-party attribution and verification) around ChatGPT, which is usually what large advertisers wait for before moving budget.