The new KPIs of AI visibility

If you cannot measure it, you cannot manage it. The old SEO metrics, keyword position, organic sessions, domain authority, still tell you something, but they describe a shrinking part of the picture. Here are the metrics we actually report on, and why each one matters.

1. Prompt coverage

For a given market, what percentage of the prompts a real buyer might type generate an answer that mentions your brand? We build a per-customer prompt set, usually 200 to 2,000 prompts spanning the funnel, and run them daily across the major models. Coverage is the percentage where you show up at all.

2. Share of voice

Of the prompts where any brand is mentioned, what share belongs to you versus each competitor? This is the closest analogue to traditional market share and the most useful single number for leadership reporting.

3. Citation frequency

How often does a model link to or quote a specific URL of yours? Citation frequency tells you which pages are doing the work, and which ones to update, expand or syndicate.

4. Sentiment and accuracy

Being mentioned is not enough. We track whether the mention is positive, neutral or negative, and whether the facts are correct. A wrong answer about your pricing in ChatGPT is a sales problem you cannot afford to ignore.

Visibility without accuracy is a liability. Measure both.

5. Surface mix

Where is the visibility coming from? Google organic, AI Overviews, ChatGPT, Perplexity, Gemini, social. A healthy mix is resilient. An over-index on any one surface is a risk waiting to materialize the next time that surface changes its rules.

How to start

You do not need a platform to begin. Pick 50 prompts that matter to your business. Run them in three models. Write down who shows up. Repeat next month. You will learn more from that exercise than from a year of rank tracking.