A catalogue is not a scraped bestseller list
Candidate discovery can use permitted category data, category-level search demand, Trends, brand recognition, stable retailer availability, topical gaps, ecosystem relevance, Search Console, internal search, comparisons, and post-activation click behavior. Each observation records its source, date, confidence, expiry, and reviewer.
Review count, sponsorship, discount language, commission rate, competitor inclusion, or attractive photography cannot select a product by themselves. Bestseller language is not inferred from an internal score.
Selection score
The 100-point score covers real usefulness, problem-solving value, demand evidence, commercial intent, topical contribution, differentiation, specification verifiability, manufacturer traceability, availability stability, comparison value, price-tier contribution, replacement ecosystem, and safety confidence.
Scores below 65 are rejected or kept only as documented internal records. A score of 75 is the editorial-research priority threshold, 82 is the standalone-page threshold, and 90 is the indexation threshold. Scores route work; they do not replace editorial judgment.
Rejection and allocation
Active recalls, untraceable identity, conflicting electrical data, unsupported certification or weather claims, unsafe construction concerns, unverifiable smart compatibility, severe seller churn, duplicate rebrands, weak support, or no durable user value can block selection.
The mature allocation covers eleven lighting and seasonal categories and targets at least 1,000 final verified records after consolidation and rejection. Up to ten percent category variance requires documented demand and usefulness evidence while the final total remains at least 1,000.