Evidence first. Routing second.
LarkMetric separates the question “what fits?” from “what can be monetized?” Those are different decisions, and payout never changes product relevance.
Evidence labels
- DOCUMENTED - current official product/provider/account evidence supports the claim.
- OBSERVED - LarkMetric reproduced a behavior under a defined test protocol. We do not use this label when no hands-on test exists.
- UNKNOWN - evidence is missing, stale, contradictory or belongs to downstream underwriting.
- INFERENCE - an analyst interpretation, clearly separated from fact.
Decision sequence
- 1 - Define the user or merchant job.
- 2 - Gather the strongest current evidence.
- 3 - Resolve fit/readiness without commercial payout influence.
- 4 - Apply eligibility, GEO, rights and legal boundaries.
- 5 - Only then choose an eligible commercial route.
- 6 - Use real conversion/revenue outcomes later to optimize among already suitable routes.
AI companion comparisons
Current public AI pages are primarily documentation-led. LarkMetric does not imply weeks of hands-on testing where it has not been performed. Future OBSERVED claims must use a reproducible memory/voice/privacy test protocol and a dated test record.
Payment readiness
Objective gaps can be labelled FIX FIRST. Questions controlled by a specialist provider remain UNKNOWN. LarkMetric does not process payments, underwrite merchants or estimate an approval probability.
Corrections and freshness
Pricing, policies and capabilities change. Material claims should carry a checked date, and corrections should update the evidence rather than silently rewriting history.