Google Is Now Reporting AI Search Impressions. Here’s What It Means for Your Home Service Business.

Search Console now shows how often your pages appear inside AI Overviews and AI Mode. The new Generative AI report can’t see clicks, queries, or your city. It’s still Google’s first native measurement layer for its own AI search surfaces, and the contractors who read it right will spot visibility shifts months before the competition notices.

Your rankings held through spring. Impressions climbed. The phone still got quieter. Home service owners have described that exact pattern to us for two years, and until June no report could even show the AI half of it.

On June 3, 2026, Google launched Search Generative AI performance reports in Search Console, its first dedicated view of AI search impressions. The new reports show how often links to your pages appeared inside AI Overviews and AI Mode, broken down by page, country, device, and date, with a separate view for generative features in Discover. You won’t find clicks, click-through rate, or query data anywhere in it. Read it as a visibility diagnostic, not a lead tracker.

The Problem This Report Finally Makes Visible

Homeowners ask Google full questions now. Should they repair or replace a 12-year-old water heater. Why the furnace short cycles at 30 below. What a 200-amp panel upgrade runs. Google increasingly answers them right on the results page, and the data behind that shift is rough if you depend on the click.

Pew Research Center tracked the real browsing behavior of 900 US adults in March 2025. When an AI summary appeared, people clicked a traditional result on 8% of visits. Without one, 15%. Links inside the summaries got clicked on roughly 1% of visits, and 60% of searches opening with question words like who, what, or why triggered a summary. That’s general US browsing rather than a trades study, but question-format searches are how homeowners research repairs.

Ahrefs measured the damage from the ranking side: across 300,000 keywords, the presence of an AI Overview correlated with a 34.5% lower click-through rate for the top organic result, and the rerun with December 2025 data put the gap at 58%. Correlation rather than a controlled experiment, but the direction matches what Pew saw. Ahrefs also found 99.2% of keywords triggering AI Overviews carry informational intent, which is why your cost guides and troubleshooting posts absorb the hit first while emergency and near-me searches still mostly resolve to the Map Pack and ads.

Google, for its part, says AI Overviews now serves over 2.5 billion monthly users and AI Mode has passed one billion. Its numbers, but the direction isn’t in dispute.

The erosion concentrates at the top of your funnel, and until June you had no first-party way to tell whether Google’s AI answers were citing your pages or skipping you.

Now you can. Partially.

What Google Shipped on June 3

Two reports, both inside Search Console. The first covers generative AI features in Search: AI Overviews, the AI summary block on a standard results page, and AI Mode, Google’s full conversational search experience. The second covers generative features in Discover, and neither includes Search Labs experiments still in testing.

What you can see

  • Impressions: how often links to your pages appeared inside generative AI features
  • Pages: which URLs earned those appearances
  • Countries
  • Devices, for the Search report only
  • Dates, with hourly, daily, weekly, and monthly views

One detail matters before you build a single dashboard: these impressions already live inside your overall Performance totals. Google carved out a dedicated view of existing data, so adding the two numbers together double-counts your visibility. Agencies are going to get this wrong in client reports all summer.

What you can’t see

No clicks. No click-through rate. No queries, no average position, no citation placement, and nothing about which passage Google’s models pulled from your page. The report answers one question: which of your pages does Google show inside AI answers, and how often. It says nothing about whether anyone called you afterward.

Where to find it

Open the Performance section of Search Console and look for the Generative AI view. If it’s missing, that doesn’t mean you have zero AI visibility. Google started the rollout with a subset of properties, UK sites first, and is expanding over time. A property can also sit out because it hasn’t collected enough impressions, or because someone flipped the new opt-out toggle. Early properties show history reaching back to mid-May 2026, though Google hasn’t documented a universal start date, so record the earliest date your own property shows before building comparisons.

How Google Counts an AI Search Impression

The mechanics matter, because you’ll compare this against AI answers you’ve seen yourself and wonder why the number runs low.

A link to your page has to be shown to the user. That’s the trigger, and it comes straight from Google’s own documentation of the report. Google’s John Mueller confirmed the corollaries: an unlinked brand mention doesn’t count, and a favicon alone doesn’t either unless it links to a page on your site.

Links hidden behind an expansion count only after someone opens them, and AI Mode tucks plenty of sources behind show-more panels, so a link nobody expands never becomes an impression.

Two more counting rules round it out, both documented in Google’s help pages. Multiple links from your site inside one AI answer roll up as a single impression in the chart total, and a property that appears twice on one results page, say inside an AI Overview and again as a standard blue link, still registers one impression at the property level.

Put together: the report counts visible, qualifying links rather than every form of AI presence, so unlinked mentions and unopened panels sit outside the number. An impression proves Google’s systems surface your pages. It proves nothing about attention or leads.

The Local Catch: There’s No City-Level View

Here’s the part that matters most for a service area business, and the part most of the coverage skipped.

Country is the finest geographic cut in the report. A Calgary furnace company sees Canada. A Phoenix roofer sees the United States. There’s no city, no metro, no service-area breakdown, and with query data absent you can’t tell whether an impression came from a homeowner ten minutes from your shop or a student writing a paper three provinces away.

That limits the report as a local diagnostic. It doesn’t kill it. The trendline shows whether Google’s AI features cite you more or less over time, the Pages view shows which content gets surfaced, and Google says it will introduce additional metrics as it works with site owners. City-level geography isn’t on any announced list, so don’t pitch it in a client report. A baseline captured now beats a blank chart later.

Close the gap with tools that already see local: GBP performance reports for calls and direction requests, call tracking for phone volume by source, and GA4 for the assistant side of AI measurement. None of them ties a specific AI impression to a specific phone call, but together they cover the outcomes this report can’t. We broke down how to read the new AI Assistant channel in GA4, which covers visitors clicking through from ChatGPT, Gemini, and Claude. You need both reports, because clicks from AI Overviews land in GA4 as plain organic search, where the AI Assistant channel never sees them.

Five Moves to Make With the Report This Month

1. Pull your baseline now

Export the Pages table, screenshot the trend, and note the date range. The rollout is weeks old and the methodology will shift, so treat month one as a reference point rather than gospel. Then add AI impressions as a standing line in your monthly reporting next to rankings, Map Pack visibility, and lead volume. Reporting that ignores AI visibility in 2026 is incomplete.

2. Sort AI-visible pages by revenue, not volume

Tag every URL in the report by type: service page, location page, cost guide, troubleshooting post, case study. A drain cleaning service page with 400 AI impressions outweighs a general-interest blog post with 4,000, because one sits next to a booking decision and the other doesn’t.

3. Rework high-value pages for clarity and conversion

Cross-reference the report against your standard Performance data and build two lists. List one: commercially important pages that rank well but show zero AI impressions. Rework those for readers. Lead each section with the direct answer under a question-phrased heading, add real numbers like cost ranges and lifespans, and cut the throat-clearing. That’s good editorial practice whether or not an AI system ever quotes it, so treat any impression lift that follows as something to monitor rather than proof the format caused it. List two: pages pulling strong AI impressions with a weak conversion path. Those need a phone number on the first screen, a booking link, and internal links to the matching service page. Visibility you can’t convert is a donation to Google.

4. Run a competitor gap analysis

Search Console only shows your own domain, so it can’t tell you who’s winning the citations you’re losing. Pair it with a tool that looks outward: Ahrefs Brand Radar tracks which domains earn citations and mentions inside AI answers, and Semrush runs similar AI visibility tracking. Pull your 20 highest-revenue queries, note which competitors appear in the AI responses, then diff that list against your own AI-visible pages. Every topic where a competitor earns a citation and you don’t is a candidate page brief. Validate it against search demand and job value before you write, since these tools sample AI answers rather than record them all, and Google notes no third-party tool sees its internal systems. Spot-check the same queries in AI Mode from your own market too, because responses shift by location.

5. Publish what only your crews know

This is the move that compounds. Google’s own optimization guidance says its AI features run on the same ranking and quality systems as regular Search, and it lists what you can skip: no special AI schema, no chunking, no rewriting pages for machines, and no separate page for every question variation, which violates the scaled content abuse spam policy when the pages exist primarily to chase rankings or AI answers. What Google says it wants is unique, non-commodity content built on first-hand experience.

Generic publishers can’t produce that. You can.

The failure pattern your techs keep finding in 15-year-old furnaces. Installed-cost ranges from last quarter’s invoices. Permit timelines in your county. Before-and-after photos captioned with what the fix was and why. A content farm can’t scrape any of that, and it doubles as the E-E-A-T evidence Google’s quality guidance keeps asking for.

What to Optimize For as AI Answers More Questions

This report tells you where you already appear. The harder question is what deserves the next content dollar when AI answers the easy stuff for free. Google’s mechanics point the way: its AI features retrieve and cite pages through the same ranking systems as regular Search, and AI Mode runs fan-out queries, related searches the model generates on its own. Google’s example is a homeowner asking how to fix a weedy lawn, which quietly fans out into herbicide, chemical-free removal, and prevention searches. Your content competes across that whole cluster, not against one keyword. Four priorities follow:

  • Click worthiness. AI answers end the shallow searches, so put your budget where a homeowner still needs you after the summary: repair-or-replace evaluations, side-by-side system comparisons, sizing and configuration, anything that depends on their specific house. Tank versus tankless for a family of five in a 1970s bungalow doesn’t resolve in one AI paragraph. Build the pages that earn the click a summary can’t replace.
  • Curiosity that keeps people exploring. Single-fact pages die on an AI-first results page. Multi-faceted resources hold attention: a visual furnace troubleshooting guide, a panel upgrade cost breakdown with photos from real jobs, a worksheet that helps someone size their own system before they call. Give a reader who arrived for one answer three reasons to keep going.
  • High-value, condition-dependent decisions. AI can’t finish a quote. Roof pitch, duct layout, panel capacity, soil conditions, and local code decide real projects, and those variables live on-site. Content built around them positions your company as the required next step instead of an optional one.
  • Structured evidence AI can cite with confidence. Google grounds AI responses in retrieved pages and leans on clear business data, and we covered how Google Business Profile connects to Gemini for exactly this reason. Keep the profile complete, keep LocalBusiness schema and NAP details consistent, and publish first-party proof like licenses, project counts, and dated job photos. The companies that get cited make their expertise easy to verify.

None of this contradicts the no-special-AI-formatting guidance above. These are content and knowledge investments a human customer would thank you for, and the AI visibility rides along as a side effect of building them well.

What to Skip, What to Keep, and One Toggle to Leave Alone

Skip the special AI markup hunt, because Google says no schema trick or machine format earns AI placement. Skip the 300 thin neighborhood pages, which sit squarely inside the spam policy Google just enforced in June. Skip reporting impressions as leads, whether to your bookkeeper or to yourself. And don’t let Map Pack work slide, since emergency and high-intent searches still pay for the trucks. This report changes how you measure far more than it changes what wins.

llms.txt deserves a more careful answer than it usually gets. Google Search ignores the file, and Google’s guidance says maintaining one for other services is completely fine, with zero effect on your Google visibility either way. Other AI crawlers can read it, it takes minutes to create, and it carries no downside, so we’d still include one. Just don’t file it under Google SEO.

The same June announcement also introduced a control, now in testing, that lets a site opt out of Google’s AI features. Keep your site included. Google states that opted-out sites “will not receive traffic or impressions” from its generative AI features, and the choice carries no ranking effect anywhere else. For a lead-generation business, opting out is a pure visibility surrender.

Where Digital Shift Fits

We run local AI search optimization programs for home service companies, and we’ve published the receipts, from what AI search optimization services actually involve to the GA4 AI Assistant breakdown linked above. The Search Console report gets the same treatment: baseline captured, citation gaps mapped against competitors, and pages built to earn AI answers moved into production.

Want a straight read on where you stand? Book a strategy call with Digital Shift. We’ll pull your Generative AI report or walk you through the likely reasons it’s missing, compare your AI citations against the competitors beating you to them, and hand you a 90-day plan that connects visibility to booked jobs.

Frequently Asked Questions

Does an AI impression mean someone visited my website?

No. It means a link to your page appeared inside AI Overviews, AI Mode, or a generative feature in Discover under Google’s counting rules. The report shows no clicks and no click-through rate, so pair it with GA4 and call tracking before drawing conclusions about traffic or leads.

Can I see which questions or cities triggered my AI impressions?

Not in this report. Impressions break down by page, country, device, and date. Query data is absent and country is the finest geographic cut, which limits what a service area business can read into any single number.

Why is the Generative AI report missing from my Search Console?

Google started the rollout with a subset of properties and is expanding over time. Your site may also lack enough impressions to display, or someone with account access may have opted the property out of generative AI features. A missing report doesn’t mean zero AI visibility.

Does the report include ChatGPT, Perplexity, or Gemini?

It covers Google’s own AI surfaces only: AI Overviews, AI Mode, and generative features in Discover. Traffic from recognized standalone assistants lands in GA4 instead, where the dedicated AI Assistant channel now groups it.

Should my company opt out of Google’s AI features?

For a business that lives on inbound leads, generally no. Opting out removes your content, traffic, and impressions from AI Overviews and AI Mode without improving rankings anywhere else. You’d surrender visibility where search is growing fastest and get nothing back.

Should I still add an llms.txt file?

Yes, we would. Google Search ignores the file and says keeping one for other systems is completely fine, with no effect on Google visibility either way. Other AI crawlers can read it, it takes minutes, and nothing about it hurts you on Google or anywhere else. Just don’t count it as Google SEO work.

Sources