If Google already answers, who visits your website?

An answer in a search engine can resolve a question before the user opens a website. For publishers, that requires separating visibility from visits. For a business, it adds a third question: how much of that attention eventually produces a useful conversation?

In July 2025, Pew Research Center published an analysis of browsing by 900 US adults. It observed clicks on traditional results in 8% of visits with an AI summary, compared with 15% without one. This is an observation from that sample, not a universal rate. [Pew Research Center]

What the figures tell us, and what they do not

The browsing studied took place in March 2025; search results were retrieved in April. That gap, and the possibility that summaries change, are among the limitations Pew describes. This is not an experiment randomly assigning the same search to two groups. [Pew Research Center]

We therefore cannot turn the difference into an exact forecast for a particular website. Nor does it mean that all visits will disappear. The editorial question that interests us is how to interpret search visibility when users can find part of the answer inside the search engine.

A citation is not a visit

Imagine one page explaining a concept and another allowing someone to compare a service’s detailed terms. A short answer might satisfy the first intention; the second might require exploring options. This is an example for thinking about different needs, not a classification we have measured.

That distinction suggests a practical approach: examine impressions, clicks and subsequent actions separately. Combining them into one number can make more appearances look like more opportunities. It can also lead to writing off an audience without checking what the people who do arrive actually do.

Google’s documentation says that normal SEO practices remain relevant and that no special markup is required to appear in its AI features. That does not guarantee a page will be selected or receive any particular number of visits. [Google Search Central]

What is worth measuring on your own website

We suggest starting with questions a business can answer using its own data: which pages generate relevant enquiries, which content helps people decide and where browsing stops. Comparing periods requires taking campaign changes, seasonality and changes to the website itself into account.

An editorial test could improve a particular explanation and observe its development without promising an automatic effect on rankings. A page’s value can also lie in original examples, clear terms or information that helps someone take the next step.

Our reading of Pew is that outcomes need to be separated before conclusions are drawn. Being visible, being visited and being chosen are not the same thing. The challenge is to understand that chain through your own data while search engines change how they present answers.