Visibility analysis
Every visibility reading is reported under two denominators and five data states. All slots stay on placeholders until sampling produces real data.
Readings
Primary denominator — full question set
Sampling40 questions. Reported as a percentage only when all three values below exist.
Numerator
—
Denominator
—
Excluded
—
Layered denominator
SamplingPurchase decision axis — A+B, 16 questions
Numerator
—
Denominator
—
Excluded
—
Methodology axis — C–G, 24 questions
Numerator
—
Denominator
—
Excluded
—
Same-family note. G#40 is in the same family as A#1 and is not double-counted in the category-leaderboard share.
Data states
- No data
Sampling has not started, or produced nothing. The slot shows a placeholder — never 0%.
- Sampling
Collection is in progress. The slot shows a placeholder — never 0%.
- Failed
Collection failed. The slot shows failed — never an inferred value.
- Partial
Part of the sample is in. Reported with numerator, denominator and excluded count.
- True zero
Shown only when the denominator is valid, sampling is complete, nothing was excluded, and mentions are genuinely zero.
Stance distribution
Recommended
No dataThe answer names the brand as a recommended choice.
Mentioned
No dataThe answer mentions the brand without endorsing it.
Negative
No dataThe answer mentions the brand in a negative light.
Absent
No dataThe brand does not appear in the answer.
Denominator note. Absent is not counted into the stance denominator.
Measurement definitions
Dual denominators
Primary: all 40 questions. Layered: purchase decision axis (A+B, 16) versus methodology axis (C–G, 24). Every reading reports both denominators and always pairs numerator, denominator and excluded count — never a bare percentage.
Stance schema
Four stances — recommended, mentioned, negative, absent. Every classification is anchored to the exact evidence span in the answer text. This page documents the defined schema; the production stance pipeline is not yet running.
Separation rules
Mentions, citations and stances are never mixed into one number. Readings are segmented when the measurement mode or caliber version changes. External research figures carry study name, sample and date. Data states are never disguised.