SearchestraWhy Two AI Visibility Tools Report Different Numbers
A brand can ask two AI visibility tools for its share of voice and get two different answers, with no obvious way to tell which is right. This is not necessarily a sign that one is wrong. It usually reflects genuine methodological differences: different prompt sets, platform coverage, testing cadence and definitions. Understanding those differences is how you evaluate what you are buying.
The fragmentation problem
Many providers now sell AI visibility measurement, each using different methodologies and rubrics. There is no single agreed definition of a mention, no universal standard for what counts as a citation, and no shared framework for whether a tool's output is reliable enough to inform strategy. Two tools can measure the same brand in the same category and legitimately disagree.
Where the differences come from
| Choice | How it changes the number |
|---|---|
| Prompt set composition | Different questions surface different brands |
| Platform coverage | A brand strong on one engine, weak on another |
| Competitive set | Changes the denominator for share of voice |
| Testing cadence | Captures a different slice of non-deterministic output |
| Definitions | What counts as a mention or citation varies |
How to evaluate rather than accept
The move is not to pick the higher number but to ask what each tool discloses. If a provider cannot or will not explain its prompt construction, platform coverage and definitions, that absence is itself a signal. See what to ask a provider.
The Searchestra view
Searchestra uses a versioned, brand-neutral prompt set and reports per engine so its numbers are interpretable and stable over time. Comparability comes from transparency, not from a single blended figure.
Tools disagree because methodologies differ; evaluate what each discloses rather than trusting the higher number, and treat undisclosed methodology as a warning.
Frequently asked questions
Why do AI visibility tools disagree?
Because they use different prompt sets, platform coverage, competitive sets, cadences and definitions. Two tools can measure the same brand and legitimately differ.
Which tool should I trust?
The one that discloses its methodology. Do not pick the higher number; pick the one you can interpret and reproduce.
Is there a standard definition of a mention?
Not universally. The lack of shared definitions is a core reason tools disagree, and why disclosure matters.