How to tell whether an AI answer cited you

Clients ask a fair question: if an AI assistant recommended us, how would we even know? The honest answer is that attribution here is harder than it was for ten blue links, but it is not a black box. There are signals, and they are worth watching deliberately rather than hoping for the best.

Read the answers yourself, on a schedule

The most direct method is also the most overlooked. Take the twenty questions a buyer in your category would actually ask an assistant, and run them, on more than one engine, written down, dated. Note whether you appear, whether a competitor appears, and which page got named. Repeat it every few weeks. A spreadsheet of real answers over time tells you more than any single screenshot, because what you care about is the trend, not a lucky day.

Watch the referrers that do exist

Assistants that cite sources send some traffic through those links, and that traffic shows up in your logs with its own referrers. It will not be the whole picture, because plenty of people read the answer and never click. But a steady trickle of visits from an assistant’s domains, landing on the exact pages you wrote to be quotable, is evidence the work is landing. Tag those pages and watch them as a group.

Separate the two questions you are really asking

“Are we cited” and “did being cited matter” are different, and conflating them leads to bad decisions. Citation is a visibility question, answered by the monitoring above. Impact is a business question, answered the way it always was: branded search lifts, direct visits, and people telling your sales team where they heard about you. If assistants are naming you, the downstream signals move even when the click path is invisible. Measure both, and do not kill a program because the middle of the funnel went dark. The middle of the funnel was always partly dark.