Data Collection, Classification & Publication Framework

Every phone record page on Reverseau is built from the same pipeline. Community reports enter the system anonymously, pass through normalisation and de-duplication, and are evaluated against threshold rules before any classification is published.

The pages below document that pipeline end-to-end - so that contributors, researchers, and regulators can verify what the data represents and where its boundaries lie. Reverseau does not perform caller identity verification and does not make determinations of legal wrongdoing.

Key Methodology Decisions

• Minimum report threshold: A number requires at least 3 independent community reports before any classification is published - enough to suggest a pattern beyond a single frustrated caller, while keeping the barrier low enough that genuine scam numbers surface quickly.
• Confidence tiers: Classifications range from "Low Activity" to "High Risk" based on report volume, consensus level, and recency weighting.
• Moderation: Reports pass through automated abuse screening (rate limiting, duplicate detection, pattern analysis) before entering the dataset.
• Data retention: Reports older than 24 months are downweighted. Number reallocation cycles in Australia typically run 12–18 months, so a 24-month window captures the full lifecycle while letting stale signals fade.
• No identity claims: ACMA allocation data tells us the service type and original carrier - not who currently holds or uses the number.

In simple terms: we collect anonymous reports, check them for abuse, count them, and only publish a safety classification when enough independent reports agree.