Google has added more ways to measure discovery: Search Console reports for website performance, a limited-rollout report for generative AI features, platform properties for eligible social and video accounts, and new Google Analytics classifications for AI-assistant referrals and source groups. That sounds like one better dashboard. For a building-industry operator, it is actually a reason to keep several lanes separate first.

A remodeler can now see a website page, an Instagram project post, a YouTube explainer, an AI-generated answer, and a later branded visit all contribute to one buyer's research. If those signals are combined too early, the report may look impressive while hiding the only question that matters: which public proof helped create a qualified call, consultation, showroom visit, estimate request, or sale?

Four reports are answering four different questions

Standard Search Console performance reports show impressions, clicks, queries, and pages that send people to the verified website property. Google's newer platform properties can report Google Search and Discover performance for verified Instagram, TikTok, X, and YouTube accounts as rollout allows. Those platform clicks are useful, but they are not website sessions.

Google's generative AI performance report, also rolling out to a subset of properties, covers impressions from AI Overviews and AI Mode. Google Analytics has separately added an AI Assistants channel for referrals from recognized assistants such as ChatGPT, Gemini, Copilot, and Grok; Google says that channel excludes its own AI Overviews and AI Mode. Each lane has a different scope, so adding the numbers together does not produce one clean visibility total.

The building-industry scorecard should start with buyer jobs

Organize the report around the work a buyer is trying to do. A homeowner may be comparing remodel approaches, checking whether a firm serves a neighborhood, looking for realistic budget guidance, or deciding whether the team understands a difficult material. A builder may be checking a supplier's availability, territory, lead times, and technical support. A designer may be evaluating process, taste, trade coordination, and documentation.

For each buyer job, keep separate rows for website Search impressions and clicks, platform-post impressions and clicks, AI-feature impressions where available, assistant referrals, branded and direct demand, and qualified business outcomes. Then annotate which pages, videos, posts, project proof, or source notes appeared. The scorecard becomes a decision tool instead of a pile of traffic.

AI answer quality needs human review

A visibility report cannot tell you whether an AI system described the business correctly. Run a small, fixed set of buyer questions every week and review the answers. Record whether Datum or the client appeared, which sources were cited, whether the source supported the claim, which caveats disappeared, which competitor appeared, and what next action the answer recommended.

This is especially important for building businesses because a wrong answer about service area, price range, lead time, installation responsibility, material suitability, or project scope can create a bad lead even when the brand received a mention. Visibility without claim fidelity is not a win.

There is still no special AI schema shortcut

Google's guidance remains straightforward: the same SEO fundamentals apply to AI Overviews and AI Mode, and there are no additional technical requirements or special optimizations required. Pages should be crawlable, indexed, eligible for snippets, helpful, reliable, and people-first. Structured data should accurately match what a visitor can see.

For Datum Field Notes, that means visible sources, a real building-industry interpretation, a canonical URL, accurate Article and BreadcrumbList data, useful internal links, and a clear next step. For a contractor or supplier, it means maintaining the same truth across the service page, project proof, Business Profile, product data, platform post, and sales follow-up.

A practical weekly operating rhythm

  • Choose five to ten buyer questions tied to real services, products, or project decisions.
  • Review website Search, platform Search, AI-feature visibility, and assistant referrals in separate lanes.
  • Check AI answers for source support, missing caveats, false claims, and the recommended next action.
  • Connect each lane to qualified calls, consults, showroom actions, estimate requests, and CRM outcomes.
  • Change one source page, proof asset, internal link, or CTA based on the evidence, then review it again next week.

The goal is not to prove that AI Search is sending traffic. The goal is to learn which sources help the right buyer make the next decision. Separate the lanes, preserve the evidence, and join them only at the business outcome.

Sources Read

Next step, if this note maps to a problem on your desk: Discovery Call — a 1-on-1 leverage assessment for your business ($1,500 · 90 min).

Related Field Notes