Measurement Interpretation
How to interpret the VRI Vorentus Recognition Index, repeated observations, manual consumer-surface capture and evidence portability without claiming more than the measurement supports.
What the VRI measures
The VRI Vorentus Recognition Index measures how well an organisation is understood, considered and chosen by AI systems under a defined measurement protocol. The current instrument separates Understanding, Consideration, Commercial Decision and Proof & Risk rather than treating visibility or mention frequency as the index itself.
Commercial Decision carries the greatest weight because being recognised or understood is not equivalent to being selected when an AI system must recommend, rank or choose.
VRI is primarily longitudinal, not a universal company ranking
The VRI Vorentus Recognition Index is designed primarily to measure the same organisation over time within a defined measurement protocol. Organisations can have different verified truth sets, markets, locales, observation classes and measurement instruments, so two identical VRI values from unrelated organisations do not automatically represent equivalent machine understanding.
VRI results are comparable only when market, locale, observation class, instrument version and measurement conditions are methodologically equivalent. Competitive comparisons are reported only where the underlying measurement design supports them. Vorentus does not present VRI as a universal public league table between unrelated organisations.
Verified Truth is the reference layer, not a score to inflate
The Verified Truth Graph establishes the organisational facts that can be supported with evidence. Those facts provide the reference against which machine understanding is evaluated; adding more claims does not mechanically increase the VRI.
A change to the measurement contract or scoring method creates a new version rather than silently rewriting historical results.
Why the Recognition Baseline uses three controlled repetitions
Three controlled repetitions are the standard Recognition Baseline protocol for repeated observation of the same experimental cell. Their purpose is to expose obvious instability while keeping the Baseline a practical diagnostic measurement.
Vorentus does not claim that three repetitions provide a classical statistical estimate of an AI system's true probability of producing a particular answer. Where variance or the decision context warrants deeper observation, Recognition Monitor or a scoped programme can use additional repeated observations.
Model changes and comparability
Vorentus records the served model or version and relevant observation conditions where available. A later retest should only replace a current comparable measurement when the applicable methodology considers the conditions comparable.
A non-comparable observation remains evidence in the longitudinal record; it should not silently rewrite the last comparable measurement.
Observation classes remain separate
Model-only observations, grounded-search observations and consumer-surface observations are methodologically different and are preserved as separate observation classes. Vorentus does not silently average them into one undifferentiated profile.
Where a consumer AI surface cannot be observed programmatically, Vorentus uses Controlled Manual Capture and reports that observation class separately.
Controlled Manual Capture: provenance is not inter-observer reliability
Operator identity, timestamp, scenario or prompt, market or locale and available evidence metadata are preserved with a Controlled Manual Capture observation.
This establishes provenance. It does not, by itself, establish a published inter-observer reliability rate, and Vorentus does not claim one where it has not been measured and documented.
Evidence portability
Authorised organisation owners can export a portable ZIP package containing organisation data in JSON and evidence files. Infrastructure credentials and security secrets are excluded.
Customer evidence portability is distinct from publication of proprietary methodology or internal APIs: the evidence supporting an organisation's measurement should remain portable without requiring Vorentus to expose infrastructure secrets or proprietary implementation details.
Interpret the measurement within its documented protocol. No score should imply more certainty or comparability than the underlying evidence supports.