How to Measure ChatGPT Ads ROI

Your dashboard says your ChatGPT ads got 40,000 impressions and 300 clicks. Your CFO asks what the ads produced. If you answered with click-through rate, you already lost the meeting — because the real question is what those ads did to pipeline, and the platform will not tell you.

Why can't you see conversions from ChatGPT ads?

Because the journey doesn't end where the ad runs. On Google, a user searches, clicks, and lands on your site — one session, one thread of tracking. Inside ChatGPT, a user sees your ad mid-conversation, keeps chatting, maybe compares you against two competitors in the same thread, then closes the app. Three days later they type your URL directly or Google your brand name. Your analytics credits 'direct' or 'organic search.' The ad that started it all gets nothing.

This is not a missing feature — it's a structural property of conversational advertising. People don't click out of conversations the way they click out of search results. They finish the conversation first. Influence and conversion happen in different sessions, on different devices, sometimes weeks apart.

Google and Meta took a decade to build this infrastructure. ChatGPT ads are at day one. Waiting for the platform to close the gap means making spend decisions blind for quarters. So you close it yourself.

What is 'dark influence' — and why does it matter more than clicks?

Dark influence is the value your ad creates without a click: the user reads your name inside a trusted AI answer at the exact moment they're evaluating your category, forms an opinion, and converts later through an untrackable path. It's the same mechanism as podcast ads and billboards — except it fires at the moment of active purchase consideration, which makes it far more potent.

Early data supports treating this seriously. Criteo reported in March 2026 that users referred from LLM platforms like ChatGPT convert at roughly 1.5x the rate of other referral channels — they arrive pre-educated by the conversation, with objections already handled. The click you can track is outperforming; the influence you can't track is likely larger. Criteo's announcement

The user who never clicked your ChatGPT ad but Googled your brand three days later is your most valuable conversion — and your attribution model gives it to organic search.

How do you measure ChatGPT ads with tools you already have?

Seven steps, in order of implementation effort. You can have the first four running this week.

1. Build dedicated landing pages per intent cluster — one page per conversation category you target, linked from nowhere else on your site. Any traffic they get is ChatGPT traffic by definition.
2. Lock a UTM convention and never deviate: utm_source=chatgpt, utm_medium=paid_ai, utm_campaign=[intent cluster]. One sloppy month of tagging destroys your baseline.
3. Track branded search volume weekly in Google Search Console. A sustained lift in searches for your brand name after campaign launch is your clearest dark-influence signal.
4. Add 'How did you hear about us?' to your signup and demo forms, with 'ChatGPT / AI assistant' as an explicit option. Self-reported attribution catches what pixels can't.
5. Extend your attribution windows. A 7-day click window misses conversions that originate from a conversation revisited two weeks later. Use 28-day minimums before judging performance.
6. Watch direct traffic to your homepage and pricing page on campaign days versus baseline. Spikes that match flight dates are signal, not coincidence.
7. Compute a blended number: total pipeline influenced (self-reported + isolated landing pages + branded-search-attributed) divided by total spend. Report that to your CFO, not CTR.

The pattern: you stop asking the platform what happened and start building your own evidence trail. This is the same discipline direct-response brands applied to podcast and TV for years — holdout logic, branded search lift, and self-reported attribution — now applied to a channel where the click is the least important event.

What should you actually report to your CFO?

Notice what's absent: impressions and CTR. Those are delivery metrics — proof the ads ran, not proof they worked. Every platform in its infancy over-indexes on delivery because that's all it can measure. Don't let the dashboard set the agenda.

Why does organic presence in AI answers multiply paid results?

Here's the compounding move most advertisers miss: the user who sees your ChatGPT ad often asks the AI follow-up questions about your category — in the same thread. If your brand already shows up organically in AI answers because you've published genuinely useful content about your space, the paid impression and the organic citation reinforce each other. The ad creates awareness; the organic presence closes the evaluation.

This is why we treat AI visibility as an extension of founder-led content, not a separate channel. The companies measuring ChatGPT ads most accurately are the ones whose content already trains the answers — they see the lift in branded search because they built the surface area to receive it. More on that mechanism here: the AI agent demand playbook · why AI-made content commoditizes

Frequently asked questions

How do I measure ROI on ChatGPT ads?

Combine five sources: dedicated landing pages with strict UTMs, branded search volume trends, self-reported attribution in your forms, extended 28-day attribution windows, and a blended cost-per-opportunity calculation from your CRM. No single source is sufficient; together they triangulate real impact.

What metrics does ChatGPT ads report?

As of 2026: impressions and clicks only. There is no conversion tracking, query-level reporting, view-through attribution, or ROAS calculation. You must build measurement infrastructure outside the platform.

Why do ChatGPT ads convert better than other channels?

Early Criteo data shows ChatGPT-referred users convert at roughly 1.5x the rate of other referral channels. By the time someone clicks, they've already discussed the category with the AI — they arrive with context, intent, and fewer objections than a cold click.

What is dark influence in AI advertising?

Dark influence is brand impact created without a click: a user absorbs your name inside an AI answer during active evaluation, then converts later through direct traffic or branded search. It's unmeasurable by pixels but shows up in branded search lift and self-reported attribution.

How long should my attribution window be for ChatGPT ads?

Minimum 28 days. ChatGPT conversations persist and get revisited, so the gap between ad exposure and conversion routinely exceeds the 7-day windows most teams default to. Judge early campaigns on branded search lift, not immediate conversions.

Should B2B companies advertise on ChatGPT?

If your buyers research with AI assistants — increasingly true in B2B — yes, but treat it as a demand-creation channel measured by branded search and pipeline influence, not a direct-response channel measured by last click. Budget accordingly.

Sources

Criteo press release — Criteo joins OpenAI's advertising pilot in ChatGPT, including the ~1.5x conversion-rate observation for LLM-referred users across Criteo's U.S. clients (March 2026). criteo.com/news/press-releases

PPC Land — coverage of the Criteo–OpenAI integration and the ChatGPT ad pilot timeline. ppc.land

EMARKETER — analysis of what the Criteo partnership means for ChatGPT's ad business. emarketer.com