How We Verify Claims
Every pain point on ConsumerLens is traceable to a public source. Here's how we find, verify, and rate the information we publish.
Every pain point on ConsumerLens is traceable to a public source. Here's how we find, verify, and rate the information we publish.
Not all sources are equal. We classify every source into three tiers:
These are official, verifiable public records. When we cite an L1 source, the claim is based on documented government action.
Examples:
These are publications with editorial teams and fact-checking processes. We require at least two independent L2 sources to corroborate a pain point before we publish it.
Examples:
These are individual consumer reports and community discussions. They can reveal emerging patterns before official sources catch up, but they are not independently verified. We cite L3 sources as signals, not as confirmed facts, and we flag them when we cannot cross-verify with L1 or L2 sources.
Examples:
For each product category, we collect consumer feedback from multiple public sources — government recall databases, review platforms, industry forums, and media coverage. We target 200-500+ individual consumer reports per category.
AI models classify each report by sentiment (complaint / neutral / praise) and by topic. Complaints are clustered into recurring themes — "battery failure" appearing in 47 reports, "warranty denial" in 23 reports, etc.
Each pain point is checked against multiple sources. A pattern that appears in both CPSC records and community discussions is stronger than one that appears in a single forum thread. We flag gaps where cross-verification is incomplete.
Every report undergoes human review before publication. We check source links, verify that quotes are accurate, and ensure our language reflects the level of certainty the sources support.
Each published page is scored on source quality metrics: L1 source coverage, cross-verification rate, and whether any claims rely on a single unverified source. Pages that fall below our quality threshold are flagged for revision.
We're surfacing patterns, not conducting scientific research. Our reports are based on publicly available consumer feedback and government records. They reflect what people are reporting — not necessarily the objective failure rate of every product. A product with more complaints may simply be more popular. We present the patterns; you draw your own conclusions.
Our source coverage is stronger in English and Chinese markets. Consumer experiences in other languages and regions may be underrepresented.