The easiest way to track AI traffic to your website.

Every founder we speak to has heard that AI search is changing how customers find brands. Very few of them can tell us whether it is happening to their business. The conversation stays at the level of opinion: someone read a report, someone else saw a LinkedIn post, and the decision to invest in content gets made on vibes.

You do not need a new tool to settle this. Three products you almost certainly already have, or can set up in an afternoon, will tell you whether AI is recommending your pages and whether those recommendations turn into real visitors.

1. 01

   Google Search Console: Generative AI report

   Open Search Console, go to Performance, then Search results, then Generative AI. You will see every page Google has surfaced inside AI Overviews and AI Mode, with impressions and clicks. Do the same under Discover if it is available for your property. This is the cleanest picture available of what Google's AI is willing to recommend from your site.

2. 02

   Bing Webmaster Tools: grounding queries

   Bing powers Copilot and shares infrastructure with several assistants, and it is the only major platform that shows you grounding queries: the query the model actually used to retrieve your page, alongside the user intent and topic. It is the closest signal you can get to the prompt behind the citation.

3. 03

   GA4: is AI actually sending visitors?

   Open Google Analytics, go to Acquisition, then Traffic acquisition, switch the dimension to Source / Medium and search for AI-Assistant. GA4 now groups assistant referrals, so you can see chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com and claude.ai as distinct sources with real user counts.

The catch nobody mentions.

Google does not show you the exact prompts people typed. Neither does any assistant. So when a tool claims it knows every AI prompt your site ranks for, take that with a grain of salt. What you can measure is impressions, citations, grounding queries and referral traffic. That is enough to make good decisions and far more than most brands in this category are working with.

What to do with the data.

Once the three reports are open in front of you, the operating decisions get obvious. If a page is pulling AI impressions but very few referral sessions, the page is being used as a source and not as a destination: tighten the opening, make the specific answer clearly yours, and give the reader a reason to come to the site rather than accept the summary.

If a grounding query shows commercial intent and you are getting cited on it, that is a page worth expanding into a proper cluster. If a whole category of queries relevant to your business shows no citations at all, you have found the content gap. Build there first.

Stop guessing whether AI is helping your business. Check the data, then build more of the content AI already recommends.

Set a recurring 30 minute block once a month to run this check. It takes less time than a single content meeting and it replaces an entire category of argument with evidence. That is the whole discipline: fewer opinions, more instrumentation, and a content plan built on what the machines are already telling you.