As of June 2026, Google Search Console includes a dedicated generative AI performance report. It shows impressions from AI Overviews and AI Mode, broken down by page, country, device, and date. For marketing ops teams who've been flying blind on AI search visibility, that sounds like progress.
It is. But only barely.
What the report actually contains (and what it doesn't)
An "AI impression" in Search Console means a link to your site was shown inside a generative AI feature. That's it. Google doesn't report which queries triggered the impression. No clicks, CTR, average position, citation placement, passage used, or conversion data. You get a count of how often your URLs appeared, sliced by page, country, device, and date.
Compare that to traditional Search Console reporting, where you get impressions, clicks, CTR, average position, and query data. The AI report is a visibility signal stripped of engagement metrics that indicate whether visibility translates to anything downstream.
The counting logic matters too. At the property level, if multiple URLs from your domain appear in a single AI response, that counts as one impression. At the page level, each URL gets its own impression. Your totals will differ depending on how you aggregate, and if you're not explicit about which level you're reading, your dashboards will mislead.
The diagnostic patterns worth watching
The report's value isn't in the raw numbers. It's in the delta between your AI visibility and your organic visibility. Three patterns are worth investigating:
High organic visibility, low AI visibility. Your pages rank well in traditional results but rarely get cited in AI answers. This suggests your content may not be structured for synthesis. Think: missing summaries, buried conclusions, or thin entity signals. The fix is structural, not promotional.
Modest organic visibility, high AI visibility. Some pages perform better in AI surfaces than their organic ranking suggests. That's a signal worth reverse-engineering. What do these pages have that others don't? Often it's specificity: a genuinely new data point, a clear framework, or a direct answer to a question that other sources hedge on.
Impression concentration. If five pages account for 80% of your AI impressions, you've got a narrow citation surface. That's a content strategy input, not just a reporting curiosity.
Connecting impressions to pipeline (because Google won't do it for you)
Here's the operational problem: AI impressions live in Search Console. Pipeline lives in your CRM. Google gives you no bridge between the two, so you have to build one.
A practical workflow:
- Export AI-visible URLs with a meaningful date range (multi-week minimum; daily fluctuations are noise).
- Pull conventional search data for the same URLs and dates. Now you can compare generative vs. organic visibility side by side.
- Categorize pages by type, topic, and intent stage.
- In GA4, isolate AI referral traffic where identifiable and map it to on-site behavior and conversions.
- In your CRM or attribution tool, look for assisted conversions on the same pages. This is directional, not definitive, but it's better than treating impressions as a standalone KPI.
The trade-off you're accepting: this workflow requires manual stitching. Google hasn't built the plumbing, and no third-party tool has reliable access to AI impression data at the query level yet. Budget time for the analysis or you'll end up with a dashboard that looks busy but says nothing.
What not to put on the dashboard
Two anti-patterns are spreading. Blended totals that add AI impressions to traditional impressions create a single number that means nothing. Attributing revenue to AI impressions without click or conversion evidence is fabrication dressed as reporting. Both will erode trust with leadership faster than having no AI data at all.
What belongs on the dashboard: AI impressions trended over time, count of pages receiving AI impressions, the relationship between AI-visible pages and their organic performance, and whatever AI referral traffic you can isolate in analytics. Label everything with what it is and what it isn't.
The baseline problem nobody's talking about
One more thing. Google didn't start reporting AI impressions until June 2026. Search Console had no AI-surface data in 2023, 2024, or most of 2025. So any year-over-year comparison claiming to show AI visibility growth from 2023 is working with numbers that don't exist in GSC. If someone on your team (or your agency) presents a "2023 baseline" for AI search, ask where the data came from. If the answer is Search Console, it's wrong.
The AI impression report is a diagnostic instrument, not a scoreboard. It tells you where Google is using your content in AI features. It doesn't tell you whether that usage drives traffic, pipeline, or revenue. Teams that treat it as a starting input for deeper analysis will gain value. Those that screenshot the trend line and paste it into a board deck will get exactly what they deserve: a number that looks good and explains nothing.
