How the AI Visibility Score is calculated.
Many tools show a number and keep the formula private. Here is ours in full, including the parts we're not satisfied with.
Mention rate
30%The share of answers that name your brand at all.
If you're not in the answer, the other inputs barely matter, so this one carries the most weight.
Share of voice
20%Your mentions as a share of every brand mentioned across the same answers.
Being named is worth less when the engine names four rivals in the same breath.
Citation rate
15%How often a mention of you also links to you.
A mention builds awareness. A citation can send a reader to your site, and it is the only part you can see in analytics.
Average position
10%Where you appear in the list of brands. Rank 1 scores 100; rank 6 or worse scores near 0.
Readers stop early. Being named fifth is closer to not being named than to being named first.
Sentiment
10%Whether the sentences that name you are positive, neutral or negative.
A negative mention doesn't help you. It is read only from sentences that name you, so praise for a rival doesn't count.
Source authority
10%The authority of the domains engines cite in those answers.
Answers built on strong sources are harder to change, so this shows what you're up against.
Website AI readiness
5%How readable your site is to AI crawlers.
Weighted lowest because it is a precondition for the others, and today it is the least directly measured input.
Each input is scored from 0 to 100, multiplied by its weight, and summed. These percentages are read directly from the code that computes your score, so if we change the formula, this page changes with it.
How an opportunity gets ranked.
Reach × commercial intent × visibility gap × confidence × ease. Each factor runs from 0 to 1, the factors are multiplied, and the result is shown on a scale of 0 to 100.
The share of your tracked engines where the gap actually shows. Measured, not estimated.
Read from the prompt's intent: commercial 1.0, comparison 0.85, informational 0.45, navigational 0.3.
How absent you are. If an engine never names you on that prompt, this is 1.
Falls with fewer observations, and falls hard when answers came from the mock provider instead of a live engine.
The inverse of effort: low 1.0, medium 0.6, high 0.3. A cheap win outranks an expensive one of equal value.
Because the factors multiply, any one of them at zero makes the score zero. That is intentional: if you can't act on an opportunity or can't trust it, it should not rank.
What this score is not
It covers your prompts only
It reflects your presence inside the prompts you chose. A different prompt set gives a different score. Pick prompts your buyers actually ask, or the score will be precise about the wrong thing.
Not comparable between brands
Two brands tracking different prompts don't have comparable scores. Compare yourself to your own last run, and to rivals inside the same prompt set.
Sentiment is keyword-based
It reads the sentences that name you and looks for known positive and negative wording. It doesn't understand sarcasm or context. That's why we always show you the sentences.
Website AI readiness needs a technical audit
This 5% input comes from the technical audit of your site (AI crawler access, robots.txt, rendering, schema). Until you run that audit, the input uses a neutral value instead of full marks.
Impact is before and after, which does not prove cause
We compare runs before and after you ship. A competitor's move or an engine update can shift the same numbers. Treat a rise as a sign you are on the right track, not as proof.