Observed, not inferred
A number appears only when it is actually visible. Nothing is back-filled from an assumption, and a missing field is never counted as zero.
METHODOLOGY · PUBLIC DATA
ClipAlytics turns public creator-marketplace observations into seller-focused direction signals. The scores help creators decide what deserves more research; they do not estimate private revenue, guarantee sales or identify individual buyers.
Method version 1.1 · last reviewed August 22, 2026 · daily UTC snapshots
01 · DATA SCOPE
Market reports describe what anyone can see without an account: how many listings compete in a category, how their prices cluster, and how much the rankings move week to week. They do not describe what any creator earned. Where a figure is not publicly available, the report leaves it blank instead of filling it with a guess.
A number appears only when it is actually visible. Nothing is back-filled from an assumption, and a missing field is never counted as zero.
Marketplaces publish different things, so some reports carry price and engagement while others carry supply and cadence only. Each report shows what it is based on.
Catalogue and sales figures are added by the creator — a forwarded receipt, a CSV, or a connection you authorize and can revoke.
02 · CLIP SCORE
Attention is a robust z-score of log-scaled social mentions, authors, and engagement versus the niche's own 30-to-90-day median. Supply growth is the same treatment of day-over-day listing change. Saturation headroom scales the gap so a crowded category needs a larger attention lead to score equally. The opportunity score is:
Opportunity = (AttentionZ − SupplyGrowthZ) × (1 + headroom)
Clip Score = clamp(50 + Opportunity × 12, 0, 100)
Headroom is (100 − saturation) / 100. Bluesky and X start at 60/40 and are renormalized when a source is off. High attention with slower supply growth pushes the score up. Rapid supply growth without matching demand pushes it down.
03 · DEMAND INDEX
The Demand Index is the Clip Score pulled toward 50 until 30 days of history accumulate. Daily social attention uses log-scaled mentions (50%), unique authors (40%) and engagement gained (10%), with a per-author daily mention cap. Each input becomes a robust z-score: (value − median) / (1.4826 × MAD). Bluesky and X begin at 60/40 weights and are renormalized when a source is disabled.
Each input becomes a robust z-score using the historical median and median absolute deviation, reducing the effect of one viral spike.
Attention relative to supply growth is adjusted by saturation headroom, so a crowded category needs stronger demand to score equally.
New series are pulled toward the neutral score of 50 until 30 days of history accumulate. Coverage records the enabled sources and confidence.
04 · DEFINITIONS
05 · CITATION & LIMITS
A defensible citation names the metric, niche, visible snapshot date and report URL. Say “public listings” rather than “sales,” and “tracked market price” rather than “earnings.” Coverage varies by source and geography, and public pages cannot reveal private purchase volume, refunds or subscriber conversion.