When we show a contractor their first AI visibility dashboard, the reaction is usually the same: "I had no idea this was being measured at all." It mostly isn't — which is why the companies that do measure it have such an edge. Here are the four numbers that describe your standing in AI answers, what each one actually means, and the mistakes people make reading them.
What is an AI visibility score?
Take every buying question that matters in your market. Ask each one repeatedly, on every engine your customers use. Visibility is the percentage of those answers that mention your company. If we run 20 questions dozens of times each and you appear in 46% of the answers, your visibility is 46%.
Why a frequency and not a yes/no? Because the engines are probabilistic — the same question produces different answers run to run. A single search is one draw from a distribution. The distribution is your actual standing.
What is share of voice in AI answers?
Visibility is about you; share of voice is about the race. Across all the answers in your market, count every company mention — yours and everyone else's. Your share of voice is your percentage of that total. It's the metric that catches competitive movement: your visibility can hold steady at 46% while a competitor climbs from 8% to 14%, and share of voice is where you see them coming.
Why does your position in the answer matter?
Engines rank implicitly. The first name in the answer gets framed as the pick; the third gets "also worth considering". Average position tracks where you land in the answers that include you. It matters because customers act on the framing — and because position often improves before visibility does, making it an early signal that your content and review work is landing.
How is sentiment measured in AI answers?
Two companies can both be "visible" while getting completely different treatment: "known for fast, clean installs" versus "has mixed reviews on scheduling". Sentiment tracking scores the language engines attach to your name and — more usefully — records the exact phrases. Those phrases tell you what evidence the engine found. If it says "slow to respond", somewhere in your review record it read that, and that's the thing to fix.
46%
A visibility score: the share of answers in your market that name you
11%
A share of voice: your slice of every company mention in those answers
2.3
An average position: where you land in the answers that include you
How do you read the four metrics together?
- High visibility, low position — you're on the list but never the pick. Usually a differentiation problem: your evidence says "competent", a rival's says "the obvious choice for exactly this job".
- High position, low visibility — when you show up, you win, but you rarely show up. Usually a coverage problem: missing cost guides or city pages for whole slices of your question set.
- Visibility up, sentiment down — you're getting named more but described worse. Check review recency; engines may be quoting an old complaint because nothing fresher exists.
- Flat everything while share of voice shifts — the market is moving around you. Find whose share is climbing and read the sources behind their answers.
Why track each engine separately?
A blended "AI score" hides the most actionable information you have. The same contractor can appear 3.5× more often on one engine than another, because each engine trusts different sources. Track visibility, share of voice, position, and sentiment per engine — the gaps between engines tell you which evidence layer needs work.
