The score
One number,
made of eight things
you can each recount.
GetMOW measures how often, where and how a company appears in the answers AI engines give buyers in its category. Eight components are read from the answers themselves and combined into one score from 0 to 100. The components are listed here. The weights are not.
The eight components
Every component is a question put to the sampled answers, and every one of them can be recounted from the logged runs. Seven are read from the answers. The eighth, retrievability, is read from the company’s own site, because an engine that cannot reach a page cannot cite it.
The eight combine into one number from 0 to 100. The weighting is GetMOW’s own and is not published, for the reason a lab does not publish its assay: the number is only worth something if it cannot be tuned to by the people it describes.
- Presence
- Does the company appear in the answer at all, named or cited.
- Share of voice
- What share of the category’s attention is the company’s, measured against the named competitor set.
- Retrievability
- Can a crawler reach, render and parse the company’s own pages in the first place.
- Citation strength
- When the company is named, are its own pages the ones cited, or is it named while another site is cited.
- Prominence
- Where in the answer the name lands. The top of an answer counts for more than a mention at the end.
- Sentiment and recommendation
- Is the tone positive, and is the mention a real recommendation rather than a passing reference.
- Engine breadth
- On how many of the engines in scope the company clears a presence threshold.
- Trend
- Which way the presence rate is moving over time, run over run.
Accuracy is watched, and kept out of the score
Wrong prices, dead features and claims that favour a competitor are tracked in every sampled answer and corrected at the source. They are deliberately not part of the score. A company that is absent from every answer cannot be described wrongly, and a score that rewarded fewer wrong claims would reward absence.
An inaccurate claim about a client is treated as the most urgent item on the board, because a buyer who reads it acts on it.
The engines
The engines are chosen per engagement, up to six of them, and fixed in writing before the first run. ChatGPT, Gemini, Perplexity, Claude and Google AI are the ones most clients choose, because they are the ones their buyers use. The set does not change mid-engagement; a change to it is a change to the contract.
How often the answers are read
One query is a screenshot, not a measurement. The same prompt put to the same engine comes back different from one run to the next, so a single answer says nothing about what a buyer will be told tomorrow. Every prompt in the agreed set is therefore run repeatedly, and the set itself never changes on GetMOW’s say-so.
Each prompt runs many times over, on every engine in scope, at a rate set to the category rather than to a standard schedule: one whose answers turn over weekly is read more often than one whose answers have not moved in a quarter. Where the readings settle the rate comes down; where they keep moving it holds. What that rate is for your category is agreed with you in writing before the baseline is taken. It is not published here, for the same reason the weighting is not: a schedule anyone can read is a schedule anyone can work around.
Every run is logged with every source the engine cited or fetched. A third-party tracker, in the client’s own name, reads beside GetMOW’s system, and if the two materially differ the lower reading governs.
Questions about the score
- What is an AI visibility score?
- An AI visibility score is one number, from 0 to 100, that says how often, where and how a company appears in the answers AI engines give buyers in its category. GetMOW’s score is built from eight components, each read from repeated, logged runs of one agreed prompt set: presence, share of voice, retrievability, citation strength, prominence, sentiment and recommendation, engine breadth and trend.
- Why is accuracy not part of the score?
- Accuracy is tracked separately because folding it into a visibility score would let absence look like correctness. A company that never appears in an answer has no wrong claims made about it, and a score that rewarded fewer wrong claims would reward not being there. GetMOW watches every sampled answer for wrong prices, dead features and competitor-favouring claims, and corrects them at the source, outside the score.
- How often does GetMOW sample the engines?
- GetMOW samples every prompt in the agreed set many times over rather than once, on every engine in scope, because the same prompt returns a different answer from one run to the next and a single reading proves nothing. The rate is set to how fast the category moves, and it is written into the client’s measurement agreement rather than published, because a sampling schedule is a thing a competitor can work around. Every run is logged with the sources the engine used.
- Which engines does GetMOW measure?
- GetMOW measures up to six engines per engagement, chosen with the client and fixed in writing before the first run. ChatGPT, Gemini, Perplexity, Claude and Google AI are the usual set. The list does not change during the engagement, because changing it would change what the baseline meant.
Sources
Read in September 2026.
Start with what the answers say about you.
Before you weigh any method, including this one, see the answer ChatGPT, Gemini and Perplexity give today when a buyer asks who to consider in your category. Free, written by hand, and yours whatever you do next.