What is an AI visibility score, and how is it weighted?
In short
An AI visibility score is one number summarising how often AI answers name or cite your brand across the questions you track. Keats weights it by intent: a buying answer counts three, a preparing answer two, a fact answer one. It's useful as a headline and a trend, and it always needs the per-question numbers behind it.
Keats, from PressRoom. Updated Read as markdown
Most AEO tools show a single score, and the question worth asking about any of them is what's inside it.
Keats calls its score weighted visibility. It's the share of answers that met their question's goal (named for buying and preparing questions, cited for fact questions), weighted by tier: tier 1 answers count three, tier 2 two, tier 3 one. Buying questions dominate it, because they're where the money is.
What a score hides:
- Which assistant. A rise on Perplexity and a fall on ChatGPT can cancel out.
- Which question. One big win can mask several small losses.
- Chance. A score built from a few dozen answers moves on its own; it needs its noise band.
So read the score as the headline and the trend, then go to the questions. AI visibility explains the rates it's built from.
Sources
- Keats: llms.txt, the operations and workflows for agents, read November 12, 2026