Ranking & AI Methodology and Limitations
Source and time
Google Maps results and a Google Search local pack are different datasets. VIDIK360's Maps module labels Maps observations. The organic Search module uses organic rank_group and does not substitute an absolute position or a local-pack position. Evidence records preserve source timestamps and parser/tracking context where available.
A result is a point-in-time observation through the selected provider and settings. It can differ from your personal browser because of location interpretation, device, language, personalization, source changes and processing time. A city-based AI prompt is not GPS targeting.
Owned listings and network coverage
Confirmed Place IDs take precedence over business-name matches. Name comparison normalizes case and repeated spaces; it does not use fuzzy similarity to silently assign ownership. Conflicting exact names require review. Organic ownership matches an exact configured hostname or its subdomains, not a lookalike domain.
For each query and location, the best matching owned listing determines network coverage. The query/location contributes at most once to the coverage denominator. Distinct owned listings in positions 1, 2 and 3 determine two-slot coverage and a complete top-three sweep. Duplicates or results from different queries/locations cannot create a sweep.
AI observations
Answer-text mentions, used-source citations and reviewed recommendations are separate observations. Mentioning a business does not prove that the answer recommends it, and a source list does not necessarily mean the source was used. No invented ordinal AI rank is assigned.
AI output can be incomplete, incorrect or inconsistent between runs. Review the captured answer and source evidence before using it for a business decision. Third-party content rights and commercial redistribution permissions remain under review.
Missing data and comparisons
Pending, failed, ambiguous or malformed checks are missing observations, not measured zero rankings. A valid complete check with no owned match is a genuine measured absence within the retrieved result depth. Disabled configuration and no scans yet are separate states.
Comparisons use paired valid checks with compatible tracking and parser context, including query/location and relevant business/provider/model settings. The interface exposes the paired denominator and missing or excluded observations. Changes in configuration or incomplete cohorts can make a comparison unavailable. Results from different providers, fixture runs or incompatible historical settings must not be blended into a claimed improvement.
Staging evidence limits
Synthetic fixtures and simulated test results remain explicitly labelled. A separately authorized, budget-limited provider pilot can establish the behavior and observed cost of those particular checks; it is not a general ranking-accuracy, scale, uptime or performance benchmark. Compare a result only with its actual source, timestamps and compatible configuration.
No score, trend, business result or provider success is fabricated to fill an empty dashboard. Draft configuration, disabled inputs, unverified readiness and no scans yet remain distinct from measured provider observations.
Document integrity
This SHA-256 identifies the archived document content. It is not a legal certification.
307ad415cf1838e576f3925520ec9183146a84c63711f97f5f7b23d70bea0c41