Measuring search and AI visibility

Start with the services, markets and buyer questions that matter to the business. Decide what would justify more content work, a technical fix or a different landing page. Define a qualified enquiry using information the team can actually assess. A visit, an email-link click and a confirmed sales conversation are different events. Report them separately. Otherwise a rise in attention can be mistaken for stronger demand even when the visitors are not suitable customers.

Record the priority pages, their publication or change dates and the available reporting period. Confirm which domain properties, locales and devices are included. Separate branded queries from broader service questions where the data allows it. Note tracking changes, migrations and periods with missing data before comparing results. For a new site, the first useful report may document crawl access and initial observations rather than a trend. Absence from a limited report does not establish absence from every search experience.

Search Console provides impressions, clicks and related performance dimensions with reporting limitations. Google currently includes traffic from AI Overviews and AI Mode in the Web search type. Do not label a change in all Web clicks as an AI-only gain. Review relevant pages and query groups, and state the filters used. Account for the difference between someone seeing a result and visiting the site. Preserve the report definition so a later comparison measures the same thing.

Where your analytics and consent setup permit it, inspect visits with identifiable referring sources and the pages they enter. Check whether those visitors take a useful next step. Keep unknown or missing referral information visible rather than assigning it to AI. A user may read an answer, return later through another route or contact the company directly. Record self-reported discovery when it is offered, but keep it separate from observed attribution. Neither source proves the whole customer journey.

Choose a small, stable set of questions linked to real buying decisions. Record the exact question, date, product, relevant locale and whether the response named or linked the site. Preserve enough context to explain what was observed. Repeat checks consistently when useful, and keep exploratory prompts in a separate set. Avoid turning one favorable answer into a general ranking claim. These observations can reveal missing explanations or incorrect company information even when they cannot quantify total exposure.

Use a short report with three parts: what changed on the site, what was observed and what action follows. Compare equivalent reporting windows when possible and flag seasonality, demand shifts or tracking changes. If qualified enquiries remain flat while visits grow, inspect customer fit and the landing-page journey before publishing more pages. If data is inconclusive, say so and choose the next check that could resolve the uncertainty. A useful report helps the team decide what to do next.