PRODUCT POPULARITY INTELLIGENCE / 4.0.0

Attention, with evidence.

Hype Score measures documented public attention in Shopfest’s research sample. It is separate from our editorial Shopfest Score. Neither affiliate commission nor a shopping link affects Hype Score.

Five dimensions, fixed weights

• Social 30%: independent original creators across verified exact-product posts. Contribution = 30 × min(1, log(1 + creators) / log(21)). Reposts and repeated observations do not add creators.
• Search 25%: median exact-product search index / 100 × 25. The authorized source must document its fixed comparison panel, geography and period. A search-result count or an index normalized independently per product is not comparable and must not be approved.
• Shopping 20%: median dated category rank percentile × 20, with the full list size recorded. This is ranking evidence, not units sold. Affiliate links and Shopfest clicks are not sales evidence.
• Consumer validation 15%: verified-purchase rating normalized from 1–5, multiplied by log(1 + verified reviews) / log(501), capped at one, then multiplied by 15. This is consumer feedback, not a Shopfest quality assessment.
• Velocity 10%: median measured growth across adjacent complete seven-day windows with identical source coverage and units. Positive growth up to 100% contributes up to 10 points. Zero baselines and incomplete windows remain unknown.

These are versioned editorial calibration choices, not measured population percentiles. Sources must be comparable within each dimension. Only the latest observation per independent source group enters each non-social dimension.

Missing means unknown

A total is withheld until all five dimensions are measured and there are at least three independent creators, two social platforms and three independent source groups. Missing weights are never redistributed or filled with zero. A measured zero contribution is distinct from unknown. Evidence coverage is the percentage of model weights with data, not an estimate of public coverage.

Labels and ranking

With sufficient evidence, 80–100 is High public attention, 60–79 is Building public attention, and below 60 is Measured public attention. Most Hyped requires at least 60 and a current verified shopping offer. A single editorial mention can only be Editorial attention only. Unscored products remain Limited evidence or Insufficient evidence. An empty ranking is an honest outcome.

Confidence is Medium when the full evidence gate is met, and High with at least ten independent creators and five source groups. Otherwise it is Low. Rising means measured growth above 10%, Cooling below −10%, and Steady within that range. Missing comparable history means Unknown.

Review, citations and freshness

Exact identity includes product ID, brand, model/name, variant and merchant SKU. Reviewers verify original IDs, source independence and authorization. Sponsored, affiliate, brand, suspicious and unknown-disclosure observations are excluded from scoring. Citations remain available as context. Duplicates are collapsed by original ID, canonical URL and source sample. Evidence expires 30 days after publication or seven days after observation. Reviews expire after 30 days. Refresh due appears when the oldest included observation exceeds 24 hours.

The test implementation supports reviewed imports and hourly recalculation. Automatic public-data collection and an hourly schedule are not connected. Recalculation never changes observation timestamps or pretends to collect new data.