Heat Score
Heat Score is a deterministic 0–100 signal for current attention relative to a site’s own history. Five explainable components are normalized before they enter the formula.
Baselines and missing data
Each site receives a persisted baseline built from accepted historical aggregates. The engine first looks for the same weekday and hour, then the same hour, then a recent rolling window. Median and median absolute deviation keep one anomalous spike from redefining normal.
- New sites show building baseline until there is enough history.
- Missing engagement is shown as unavailable and receives a neutral subscore with lower confidence.
- Historical records remain when a tracker becomes stale; the live ranking state changes instead.
Eligibility and score states
Score state is separate from the displayed number. A provisional score can be useful for discovery while history and confidence grow, but it is capped and kept distinct from fully eligible organic ranking.
Leagues
New, Emerging, and Established leagues compare sites with a more appropriate amount of history and scale. League changes use hysteresis so a site does not bounce between labels because of one update.
League membership never changes the underlying formula and is not a way to buy visibility. It only controls the comparison set shown to users.
Organic rank
Public pages read transactionally published current rankings. The deterministic ordering is displayed score, smoothed score, confidence, accepted visitor volume, calculation time, domain, and site ID. Rank movement comes from persisted snapshots, not a client-side guess.
- Global, league, category, category-and-league, new, and breakout scopes are separate views.
- Smoothing and breakout persistence reduce noisy oscillation while allowing large genuine changes to move quickly.
- Sponsored placements remain separate from the organic ranking record.
Breakouts
A breakout requires multiple signals: meaningful relative lift, meaningful absolute lift, live acceleration where supported, fresh accepted traffic, adequate confidence, and persistence across evaluation windows. Entry and exit thresholds are different, and cooldown prevents repeated alert spam.
Verification and sponsored reach
Traffic proof and ownership proof are separate. Tracker and GA4 labels identify the measurement source; claimed ownership does not turn unverified traffic into verified traffic.
Versioning and known limits
Every persisted score stores its version, calculation window, baseline reference, components, penalties, confidence, and state. The current release is Heat v1. Historical versions remain auditable when the formula changes.
Production quality still depends on source freshness, provider availability, baseline history, and traffic classification. The system is not bot-proof, does not claim scientific validation, and does not infer causality from a breakout.
Submit a site