AI visibility is getting easier to observe. That does not mean ROI is getting easier to prove.
In some ways, the opposite is happening.
More discovery is happening inside AI answers, across assisted journeys, and through visibility patterns that do not map neatly to the old click-first model. So the ROI question has to get sharper.
Why ROI gets messier
Google's AI-features documentation and June 3 Search Console reporting update both make the visibility layer easier to see.
Bing's June 16 reporting expansion gives more citation context too.
But TechRadar's June 23 piece on AI ROI warned against measuring only at a tool level. That applies here as well. A dashboard can show that AI visibility exists without proving that it creates meaningful business movement.
Analysis: this is why AI-search ROI is not a screenshot problem. It is a systems problem.
Signals that matter
The signals I would trust most are:
- Cited page type. Service and proof pages are usually more valuable than generic informational pages.
- Assisted conversions. Did AI-visible content show up in real buying journeys?
- Branded return demand. Are more people coming back by name after earlier AI-mediated discovery?
- Lead quality. Are inquiries better informed and closer to fit?
- Commercial topic coverage. Are you becoming visible for the questions that actually matter to revenue?
Signals that flatter you
The flattering signals are the ones that feel impressive but do not change business outcomes:
- one-off citations with no persistence;
- impressions on pages that never lead anywhere;
- visibility growth without better leads;
- citation wins on curiosity topics far away from your offer.
Those metrics are not useless. They are just not enough.
A practical model
If I were building a lean ROI model for a small brand, I would connect four layers:
- Visibility: Search Console AI reports plus Bing Citation Share and topic patterns.
- Page value: which pages are getting the visibility.
- Journey value: whether those pages assist deeper visits or conversions.
- Lead value: whether the resulting inquiries are actually better.
That is the difference between AI visibility that looks exciting and AI visibility that earns budget.
CTA: If you want to prove AI-search ROI, stop asking only whether visibility is up. Ask whether the right pages are becoming visible often enough, for the right questions, and with better business outcomes attached.
The one number I would actually report
The four-layer model is the analysis. But if someone made me report a single figure to justify AI-search work, I would not report impressions or citations. I would report the share of new, well-fit inquiries that mention or arrive through an AI-mediated source.
It is a blunt number, and that is the point. It sits at the far end of the chain, past visibility, past clicks, right next to revenue, so it cannot be inflated by curiosity traffic or one-off citations. If it moves up, the visibility work is reaching real buyers. If it stays flat while impressions climb, you have proof that the visibility is landing on the wrong pages or the wrong questions, which is a genuinely useful thing to learn early.
Getting the number is low-tech: add one line to your intake, "how did you find us," and tag the answers. It will be imperfect, because AI-mediated discovery is often invisible and buyers misremember. But a rough directional read on the metric closest to revenue beats a precise dashboard of the metrics furthest from it. The precise vanity number and the rough revenue number are not equally useful, and only one of them earns budget.
Sources
- Google Search Central: AI Features and Your Website
- Google Search Central Blog: Introducing Search Generative AI performance reports in Search Console (June 3, 2026)
- Bing Search Blog: New AI Visibility Insights in Bing Webmaster Tools (June 16, 2026)
- TechRadar Pro: Measuring AI ROI at tool level is missing the point (June 23, 2026)