3.See where you stand.

You cannot manage a reputation you have never seen. This chapter is about getting your baseline: the handful of numbers that describe where you stand in AI answers, and what each one actually tells you.

The four numbers that matter

Mention rate is presence: out of all the answers collected for your tracked questions, what fraction named you at all. It is the bluntest number and the first one to look at. A low mention rate on the questions that matter means the assistants are recommending your market without you in it.

Share of voice is how much of the room you take up: of every brand mention across those answers, the percentage that is you. Share of voice separates “we appear” from “we matter”: two brands can both show up in most answers while one leads every list and the other trails at the bottom.

Citation rate is whether your own website makes the reading list: the fraction of answers that cite your pages as a source. A decent mention rate with a near-zero citation ratemeans AI describes you entirely in other people’s words, which is exactly as comfortable as it sounds.

Authority citations are the independent sources engines lean on when they talk about your market: review platforms, industry publications, best-of lists. Which authorities get cited, and whether they mention you, is where most visibility is actually won and lost. More on that in chapter five.

Scores summarize; receipts prove

Any tool in this space will blend numbers like these into a single visibility score. Scores are useful for tracking change over time. Just never accept one without receipts: the actual answers, verbatim, that the score was computed from. If a dashboard says you lost ground, you should be able to click through to the exact answers where it happened and read them yourself. Demand this from any vendor, ours included.

Getting your baseline

A useful baseline needs three properties. It asks questions real buyers ask, in their words, not your marketing’s words. It asks every major engine, because of chapter two. And it asks enough questions to smooth out the run-to-run variance, because a single ask is a coin flip.

That is exactly what the CiteHawk free report is: 50 buyer questions generated for your specific market, asked across all 8 assistants, scored and receipted. It is the complete product run once, not a teaser, and it takes minutes to start.

Reading your first report without panicking

Three habits for the first read. Start with discovery questions, the ones where buyers do not know brands yet; that is where invisible losses live. Read a few full answers before reacting to any number; the texture of what the assistants actually say teaches more than the aggregate. And treat anything surprising as a hypothesis to confirm next collection, not an emergency: variance is real, and real movement shows up as a rate that moves and stays moved.