GeoScript Methodology / Data Freshness
Data Freshness
How GeoScript timestamps evidence, handles stale signals, and keeps published reports permanently accurate as a historical record.
Freshness timestamps on scores and reports
Every opportunity score record carries a freshness_at timestamp — the moment the score was last computed. This timestamp is visible in the product UI so that users can see how recent the underlying computation is, and whether the evidence feeding that score has aged since it was last run.
Published reports carry the report period window explicitly: a period.start and period.end with half-open interval semantics ([start, end)). These boundaries describe exactly which observations fell within the report window. A report does not describe "approximately" a time period — it describes a precisely bounded one.
How freshness affects the Opportunity Score
The Opportunity Score includes a Freshness component (5% weight). This component is scored as a freshness fraction: 1.0 means all evidence is within its registered freshness window; 0.0 means all evidence has expired.
Stale evidence lowers the Freshness component score — it is never treated as neutral or current by default. An opportunity with entirely expired evidence will receive a zero on the Freshness component, which reduces its total score. This is intentional: acting on stale evidence is a real risk, and the score should reflect it.
The Evidence Confidence component (also 5% weight) independently captures the aggregate reliability of the observations used. Low-confidence observations from old or sparse data sources pull this component down alongside Freshness.
Published reports are immutable historical snapshots
When a report is published, GeoScript records a deterministic snapshot of all rollup counts — recommendations, deployments, page coverage — at that exact moment. Once published, the report cannot be modified.
This immutability is enforced at the database level, not just as a service-layer convention. A database trigger prevents updates to published report records. New data that arrives after a report is published does not alter the report; it may appear in a future report covering a later period.
The report itself states this explicitly in its narrative: “This report is a permanent snapshot; new data appearing after publication does not modify it.”
This means a report you shared with a client three months ago describes exactly what GeoScript observed during that period. It is a permanent historical record, not a living document.
What GeoScript does not do
GeoScript does not silently recompute published reports with newer data. A published report describes its period and nothing else.
GeoScript does not treat missing or expired evidence as if it were fresh. A gap in evidence reduces the score; it is never filled with an optimistic placeholder.
GeoScript does not fabricate a freshness signal. If evidence has not been collected for a business, the Freshness component reflects that absence directly in the score.