Trust and Methodology
Ask Proof Methodology
Ask Proof is built around transparent components instead of an unexplained composite visibility score. The methodology preserves enough context to understand what was observed and what was not.
Definitions
- Buyer query: the exact rendered question sent to a surface.
- Surface: one collection channel.
- Observation: one prompt checked on one surface for one location at one timestamp.
- Evidence: raw response, rendered text, screenshot, citations, and collection metadata.
- Market context: country, language, timezone, and optional region, town, or coordinates.
Collection and normalization
Raw evidence is stored before interpretation. Deterministic extraction is preferred for target mentions, citations, domains, and entities. Provider and collection metadata are retained for later comparison.
Uncertainty is explicit
Blocked, missing, ambiguous, and failed observations remain separate states. They are not converted into non-mentions or zero scores. A partial run is reported as partial.
Comparison rules
Weekly comparison is limited to like-for-like observations. A different question, provider, model, location, collection mode, or relevant time context can change the result and must remain visible.
Claims and limitations
- One observation does not represent every customer.
- Location-neutral observations are not universal worldwide results.
- No implementation guarantees visibility improvement or an AI recommendation.
- Simple base actions are not a guarantee and detailed diagnosis is a separate entitlement.