1. Observe
For the launch beta, Apple iOS metadata, chart ranks, listing facts, and supported product signals are persisted with source and observation timestamps. Android remains unavailable until its own approved provider path exists.
Evidence over vibes
A number is only useful when you can see what supports it.
AppSignal combines public store observations with a separate estimation pipeline. We keep the observation, model output, coverage, and confidence visible so a research page never presents a modeled number as official publisher data.
2. Estimate
The existing ensemble uses legitimate rank, history, category, storefront, monetization, and rating evidence. Missing evidence remains missing.
3. Disclose
Every public estimate carries a range, confidence, model/feature version, period, drivers, limitations, and a clear gross-consumer-spend definition.
Revenue semantics
- Gross store consumer spend is separate from developer proceeds.
- Subscription MRR, web revenue, and advertising revenue are separate channels and remain null without evidence.
- Country contributions are aggregated only after country-level predictions exist.
History integrity
- Charts display actual collected dates only; a four-day history stays four days.
- Censored or missing ranks are not converted into zero performance.
- Short history and model disagreement reduce confidence and widen intervals.