How to Measure Your Website’s Visibility in Google AI Search

There is now more to inspect than a prompt test

Google’s Search Console announcement says its dedicated generative AI performance reports rolled out worldwide by August 31, 2026. They provide a view of impressions within features such as AI Overviews and AI Mode, with page, country, device and date breakdowns.

A second recent change matters for businesses with visual work. On September 24, Google announced multimodal reporting for searches using images, including Lens and Circle to Search. It is available through a search-type filter in the Search results and generative AI performance reports, with a global rollout beginning on that date.

These updates make October a useful time to review your measurement setup. They do not mean every site will have substantial data, or that appearing in an answer automatically produces business. Start with the report your property actually shows, then connect that observation to what happens on the website.

Decide which question each number answers

Before exporting anything, write down what you want to learn. For a Vancouver service business, that might be whether its service pages receive relevant exposure, whether visitors reach the portfolio and whether the inquiry path works. A single ‘AI visibility score’ would compress those questions too far.

A company may see more impressions while getting fewer inquiries. That is a signal to investigate, rather than proof that the content is succeeding or failing. The offer, audience, season, contact flow and source mix can all matter. Keep the separate observations visible.

Review one property and a consistent date window

Open the appropriate verified Search Console property and its generative AI performance report for Search. Check that the property represents the URLs you intend to study. Keep Discover separate if the question concerns website discovery through Search; the launch announcement describes a distinct report for Discover.

Google’s report documentation defines impressions as links to your site being shown within supported generative features. It also explains why the report may be absent, including insufficient impressions, and notes that Search Labs experiments are excluded. No visible report is not proof of a technical penalty.

• Record the property, date window, report and active filters.
• Review the pages receiving impressions and whether they support your actual services.
• Use country and device views to check the context of that exposure.
• Compare equal-length completed periods before investigating a change.
• Save the export and a short note about the question you were trying to answer.

The documentation says recent data can be preliminary and explains how page-level and property-level aggregation differ. Avoid adding every page row and assuming it must equal the chart total. If numbers disagree, check the reporting definition before treating the discrepancy as a business trend.

Use the multimodal filter for a specific question

For a builder, designer or hospitality business, visual discovery may be worth examining. Google’s September announcement says the new filter includes searches originating from Google Lens, Circle to Search on Android, image uploads and Chrome’s image-search action. It says metrics become visible when the site receives traffic from those queries.

A practical question is whether the pages supporting your real work appear through these searches. For example, a portfolio page with clear project context gives a visitor something useful to assess after discovery. A gallery with beautiful images but no explanation leaves the commercial meaning unclear.

Do not turn a small or empty dataset into a requirement to replace the site’s visual identity. First inspect the available evidence and the pages involved. If the website needs stronger project descriptions or genuine project photography, improve those because they help a customer evaluate the work. Our selected portfolio distinguishes project scope from atmospheric imagery; the same distinction matters when reviewing visual discovery.

Connect exposure to the customer journey carefully

Use your analytics platform to review available source, referral and landing-page information. A reported Google visit is not, by itself, evidence that it came from an AI Overview. Do not label all organic Google traffic ‘AI traffic’, or attach an individual lead to a Search Console impression.

Define a meaningful inquiry event in the implementation. A form-button click can happen without a valid message being sent. A confirmed server response is a better place to record successful delivery. Direct email and telephone inquiries may need a separate, consistent record so they are not lost from the business picture.

Then review quality: was the inquiry relevant to your service, geography and project scope? Keep personal inquiry details out of public dashboards and ordinary analytics events. Counts and a clear classification are enough for a basic operating review; customer conversations belong in the appropriate business system.

For a Vancouver website, the useful outcome is a relevant conversation, not merely a new graph. If visitors arrive but struggle to find the scope, work examples or contact method, content and interface improvements may matter more than publishing another article.

Keep a small decision log

We would start with a monthly review rather than a daily reaction to every movement. Use the same date-window convention, save the exports and annotate meaningful website changes: a rewritten service page, a new project, a repaired form or a migration. If the business has very little data, allow more time before drawing conclusions.

If a service page has exposure but its content does not help someone choose the business, review that page. If the inquiry flow fails, repair the flow. If a useful page is not indexed, investigate access and indexing separately. Avoid changing copy, navigation and measurement simultaneously when you want to learn which problem an edit addresses.

Keep AI discovery, agent usability and measurement connected without treating them as the same job. The first helps you understand how content is found; the second asks whether the interface supports a task; this review checks the evidence available afterward.

Menerai brings search foundations and clear inquiry paths into website design and development. If you are planning a rebuild, tell us about your project. We can help define what the site should communicate, how it should work and which outcomes deserve to be measured.