WHAT HAPPENED?
ChatGPT Ads is starting to get the less glamorous infrastructure that serious paid-media channels need: measurement and verification.
OpenAI announced expanded measurement partnerships alongside a new visual ad format, conversion-data integrations, early incrementality work and brand-suitability evaluation pilots with DoubleVerify and Integral Ad Science.
ChatGPT Ads already had the attention.
Now OpenAI is building more of the infrastructure advertisers need to decide whether that attention is actually worth paying for.
1. WHAT OPENAI ACTUALLY CHANGED
The announcement touches three different parts of the advertising stack:
Advertisers can also use:
MEASUREMENT
OpenAI added integrations with Hightouch, Tealium and LiveRamp for sending conversion data, alongside a wider group of web, app, full-funnel and advanced measurement partners.
BRAND SUITABILITY
OpenAI is developing evaluation pilots with DoubleVerify and IAS. These are pilots, not universal brand-suitability controls available to every advertiser.
NEW VISUAL FORMAT
OpenAI plans to test a new visual ad format with an initial group of U.S. advertisers during image generation. Ads will remain labelled and separate from the generated image.
OpenAI has also made Negative Phrases available to qualifying advertisers for narrower placement requirements.
This matters because conversational advertising creates a different verification problem.
The context isn't simply the webpage or video surrounding an ad.
The context can be the conversation itself.
ChatGPT already had the unusual new advertising environment.
What it's now acquiring is the boring stuff that makes advertisers comfortable spending more money: measurement, verification and evidence.
2. WHY IT MATTERS FOR ADVERTISERS
The important shift is simple:
ChatGPT Ads is becoming easier to judge as a media channel rather than an AI experiment.
Until now, the interesting questions were mostly about the experience.
Can ChatGPT reach someone while they're researching?
Can advertising enter the journey before traditional search?
Can a sponsored interaction influence a later decision?
Now comes the less exciting, but commercially much more important question:
Did it work?
DoubleVerify's attribution capability gives advertisers another way to incorporate ChatGPT Ads into the wider customer journey rather than viewing it only through ChatGPT's own reporting.
OpenAI is also starting to explore incrementality.
Those two concepts shouldn't be confused.
Attribution asks: Who gets credit?
Incrementality asks: What happened because we advertised?
Instead of asking whether the platform is interesting, advertisers can start asking whether it produces acceptable CPA, ROAS and customer quality.
That is when a new advertising product starts becoming a real media channel.
3. Here’s an Explainer!
Imagine a wellness brand advertising across Google, Meta and ChatGPT.
A potential customer encounters the brand while researching something inside ChatGPT.
They don't purchase.
Two days later, they Google the brand.
Then they buy directly from the website.
Who gets the credit?
Last-click measurement might make the final touchpoint look responsible for everything.
Adding ChatGPT to broader attribution measurement gives the advertiser another way to understand whether that earlier interaction contributed to the journey.
Suppose — hypothetically — the advertiser sees:
ChatGPT attributed CPA: $75
Paid Search benchmark: $85
Interesting.
But that's still not enough to justify shifting the budget.
The advertiser needs to ask:
Were these genuinely new customers?
Would they have converted anyway?
Does ChatGPT assist conversions credited elsewhere?
Does performance hold as spend increases?
There are already early partner-reported results from advertisers testing the channel, but these should be treated as individual early examples rather than performance benchmarks for ChatGPT Ads as a whole.
THE SIGNAL
Attribution can make ChatGPT easier to understand.
Incrementality will help determine whether it deserves more budget.
4. WHAT ADVERTISERS SHOULD DO
First, measure ChatGPT alongside the rest of the journey.
Where available, connect ChatGPT Ads to the measurement systems you already use rather than evaluating it solely through platform reporting.
Second, create a benchmark before scaling.
Compare CPA, new-customer rate, revenue quality and downstream customer value with established acquisition channels.
Third, separate attribution from incrementality.
Showing that ChatGPT participated in a conversion journey is useful. It doesn't automatically prove that ChatGPT created the conversion.
Fourth, verify the suitability controls actually available to you.
The DoubleVerify and IAS initiatives are still evaluation pilots. Don't treat them as fully mature, universally available controls yet.
ORBIT. TODAY’s TAKE
The flashy part of this announcement is the new visual ad format.
The strategically important part is measurement.
New ad channels don't become serious because they invent interesting placements.
They become serious when advertisers can compare them with existing channels, independently evaluate their contribution and eventually determine whether they created incremental business.
ChatGPT Ads is getting closer to that point.
The question is moving from “Can ChatGPT carry advertising?” to “Can advertisers prove ChatGPT deserves the budget?”
