How AnswerSignal separates observation, measurement quality and change.

24 neutral Discovery and 6 Brand Understanding questions.

Discovery questions test visibility across real purchase, provider and problem-solving intents without naming the brand. Brand Understanding questions test how accurately and consistently AI systems understand the business.

Questions are reviewed, then frozen.

A person assesses relevance, neutrality and business value. Official measurement starts only with exactly 30 accepted questions in a versioned, frozen set.

Five reference intents monitor measurement stability.

Selected neutral questions cover core intents. They support methodology stability without artificially expanding the official visibility metric.

Seven full measurements across four systems.

Each full measurement runs the same approved question version separately through ChatGPT, Claude, Gemini and Perplexity. Provider retries do not count as additional full measurements.

Provider failure is not zero visibility.

AnswerSignal reports completeness, coverage, provider status and comparability. Missing answers remain missing evidence. They never reduce brand visibility artificially.

Only like-for-like measurements are compared.

Comparison requires the same question version, sufficient provider coverage, complete processing and correct time order. Historical measurements remain immutable snapshots.

Recommendations require evidence and relevance.

A recommendation is eligible only with complete, same-version, comparable evidence, sufficient measurement quality and a traceable business rationale. It is not tied to a fixed Wave.

Observed change follows documented implementation.

Only a later comparable full measurement after approval and documented implementation can show what changed.

A further measurement checks persistence.

An initial positive signal may be temporary. Persistence asks whether the observed change remains in a later comparable measurement.

Research stays outside official client measurement.

Where used, Shadow Research explores variants and robustness separately from official full measurements. It changes no client metric and never appears as an extra Wave.

What AnswerSignal does not prove.

AnswerSignal does not prove causal impact from one action, future rankings, revenue or stable answers from every AI system. It observes change under documented conditions and labels uncertainty.

• ✓No ranking or revenue guarantee
• ✓No blending of provider failures with visibility
• ✓No automatic website changes
• ✓Search Console and GA4 as optional context only