Data & Methodology Audit

Continuous sub-score curves. Replaced the remaining binned look-ups (fire-history proximity, slope, fire size) with continuous interpolation, so a hair more distance or slope no longer jumps the score across a band edge.

Recency as a multiplier. Fire-history recency now scales the base (≤1.0) instead of an additive bonus — a single decades-old burn no longer reads "High" on its own, while recent fires are unchanged.

No fabricated terrain. When USGS terrain is unavailable, slope is dropped and weights renormalize — it is no longer assumed flat (which had biased scores down).

Location-precision gate. A search that resolves only to a city/ZIP centroid (not a full street address) is flagged and its confidence capped at Moderate, so a centroid is never presented as a confident parcel score.

FEMA "Insufficient Data" excluded. Counties FEMA rates "Insufficient Data"/"No Rating" are dropped rather than scored as a guessed middle value, and don’t count toward confidence.

Primary sources only. Every data point and external link resolves to a government or official agency (.gov / federal program) — the original owner of the data, not an aggregator.

No proprietary or paywalled scores. We don’t resell a third-party black-box risk number. The FireRisk is computed only from open federal datasets, with the formula and raw values published.

Dated and re-checked. Links and facts are re-verified on a schedule; this audit was last completed in June 2026. Live feeds are monitored continuously on our status page.

We show our work. The full weighting, the renormalization rule, the raw RPS value, and the confidence level appear on the report and the methodology page — anyone can audit the math.

Independent, with disclosure. FireRisk.ai is not an insurer, agent, or financial/tax advisor. Where we connect you with a partner (contractor, restoration, insurance), we disclose that we may be compensated. Editorial ratings are never sold.