Findable Methodology

What is measured

Findable measures how brands appear in the answers generated by AI assistants when those assistants are asked buyer-style questions about a category. The unit of observation is an answer: one AI system's response to one prompt at one point in time.

Findable does not measure search-engine rankings, ad placements or website traffic. It measures what AI systems say, and the sources they cite while saying it.

How observations are collected

1. A prompt set is assembled for the brand's category, covering the questions a buyer would realistically ask.
2. Each prompt is sent to the monitored AI systems and the returned answer is stored as evidence.
3. Each answer is parsed for brand mentions, competing brands, the way the brand is characterized, and the sources cited.
4. Those per-answer observations are aggregated into the scores and comparisons shown in the product.

The AI systems currently observed are ChatGPT, Claude, Gemini, Perplexity. No other answer system is included; Google AI Overviews is outside that set.

Measurement is run-based. Observations are produced when a run executes, either on request or on a schedule. Findable does not observe AI answers continuously, and figures shown in the product describe the runs that have already completed.

How the AI Visibility Score is composed

The AI Visibility Score summarizes a brand's presence across the answers observed for its prompt set. It combines three components:

The score's definition is owned by /guides/ai-visibility-score; this page documents only how the components are produced.

How sentiment is treated

Sentiment is evaluated only for answers in which the brand is actually mentioned. An answer that does not name the brand is not counted as negative sentiment — it is counted as an absence of presence, which the Presence and Coverage components already capture. Where an answer mentions the brand but expresses no evaluative view, it is treated as neutral rather than being discarded or scored as negative.

Discovery Signals and AI Readiness

Two distinct diagnostics sit alongside answer measurement. They are not the same score and are not interchangeable.

Discovery Signals — the technical audit of your own website

Discovery Signals scores a website against the technical and structural conditions AI retrieval systems depend on. Each individual check resolves to pass, partial (where a partial state is meaningful for that check) or fail; checks that could not be evaluated are excluded from their category's denominator rather than counted as failures. Category scores are then combined with fixed weights:

AI Readiness — the brand-wide composite

AI Readiness is a brand-wide composite built from four pillars: Crawlability, Mentions, Reviews and Web Presence. Discovery Signals is the deep technical audit of the brand's own website and rolls up into the Crawlability pillar. How much each pillar contributes depends on the brand's archetype, so Findable does not publish a single universal weighting table for AI Readiness. The authoritative product description is /platform/ai-readiness.

Competitive and market measurement

Competing brands are identified from the same observed answers, so a competitor set reflects who AI systems actually put in front of buyers rather than a manually curated list. Brands a team defines itself take precedence over brands detected automatically. Market coverage is measured across the prompts and topics that define a category, not against a fixed universe of competitors.

Term definitions: competitive visibility, market coverage, source authority.

Documented limitations

• Sampled, not exhaustive. Measurements describe the prompts that were tested and the runs that completed, not every possible question a buyer could ask.
• Not real-time. Observations are produced by runs. Between runs, an AI system's answers may change without that change being reflected.
• Non-deterministic sources. AI systems can return different answers to the same prompt. Repeated observation reduces, but does not remove, this variance.
• Limited system coverage. Only the four systems listed above are observed. Anything outside that set, Google AI Overviews included, is not measured.
• Diagnosis, not causation. Findable can show that a brand's presence changed and which signals are weak. It cannot prove that any single action caused a change inside an external AI system.
• No guarantees. Findable does not control external AI systems and does not guarantee mentions, citations, recommendations, rankings, traffic or revenue.

Which page owns which fact

If this page ever conflicts with the owner page for a specific fact, the owner page is the current source.

Corrections

To report an inaccuracy in this methodology, use the contact form at /contact.

Last reviewed: 2026-08-09