PolicyChat Rate Authority — Data Sourcing and Methodology

PolicyChat Rate Authority — Data Sourcing and Methodology

Effective: 2026. Maintained by: PolicyChat Editorial. Last reviewed: monthly.

PolicyChat Rate Authority publishes daily-refreshed US insurance rate data. This page documents every step from source to published page — what we ingest, how we validate, how we attribute provenance, and what each data point means.

1. Data sources

Three distinct sources feed the Rate Authority ledger. Each record in our database carries an explicit source field; every published page cites the source per row.

1.1 State Department of Insurance / SERFF filings

The System for Electronic Rate and Form Filing (SERFF), operated by the National Association of Insurance Commissioners (NAIC), is the public-domain filing system carriers use to submit rate changes for regulatory review. We ingest filings within hours of public posting.

Per state: each US Department of Insurance also publishes filing search interfaces (often pre-PRA actuarial memos in addition to SERFF). We integrate state-specific scrapers for California, Texas, and Florida directly, and use SERFF as the canonical national source for the remaining 47 states.

Provenance tag: source: state_doi_serff with the filing ID and source URL preserved on every record.

1.2 NAIC published state averages

NAIC publishes the Auto Insurance Database Report and Homeowners Insurance Report annually. These are state-aggregate baselines (not carrier-specific) covering all coverage types combined. The most recent available data is typically 18-24 months behind the publication date.

Provenance tag: source: naic_state_average_<year> with the explicit data year (NOT the publication year) on every record. We do NOT timestamp- launder these as current data.

1.3 Licensed partner feeds

PolicyChat operates licensed lead-distribution agreements with EverQuote (NASDAQ: EVER) and LendingTree (NASDAQ: TREE). These partner feeds provide real-time, per-applicant-profile quote data when a user walks through the Sage interactive product. Partner-feed records are used for personalized quoting; they are NOT aggregated back into the public Rate Authority baselines.

Provenance tag: source: partner_<provider> with the partner-issued quote ID preserved.

2. Ledger architecture

The Rate Authority ledger is an append-only SQLite database keyed by (state, carrier, product, effective_date, profile_hash). Records are immutable once written. Rate changes appear as new records, not in-place updates — this preserves the historical record for time-series analysis and gives any consumer of a published page the ability to reconcile a current figure against its source filing.

Columns: state, carrier, product, effective_date, profile_hash, monthly_premium, deductible, coverage_limits, source, source_url, ingested_at.

3. Page-level data fidelity

A published state×product page may include data from one or more of the three sources above. The fidelity declaration appears prominently above the data table:

• If only NAIC baseline is present: the page is labeled as “2023 NAIC baseline” — never as “current rates.”
• If recent SERFF filings are present: each row in the carrier table cites its filing ID + effective date. The page header reflects this.
• If partner-feed quotes exist for the profile: those are pulled live via Sage rather than embedded in the static page.

4. What “best for X” routing means

The “Best for low-rate shoppers / high-asset households / etc.” subsections on each state page are NOT carrier endorsements. They are evidence-anchored routing recommendations based on:

• The most recent filed baseline rates from our ledger
• Carrier underwriting appetite per profile class (publicly documented)
• Historical complaint ratios (NAIC Complaint Index)
• Policy structure quirks for the profile (e.g. SR-22 availability, nonstandard auto specialization)

When carrier filings move, the routing updates. We do not accept payment from carriers to be featured.

5. What we do not do

• We do not invent rates. Every dollar figure traces to a source record.
• We do not republish carrier press releases as if they were our analysis.
• We do not aggregate scraped quote-flow data from carrier websites. Quote data comes from licensed partner feeds only.
• We do not editorialize beyond what the data supports. When we offer a direct recommendation (e.g. in the Verdict Engine), the recommendation is gated by an explicit conviction tier (see /methodology/conviction-tier/).

6. Errata and corrections

If you find an error in a Rate Authority page, email errata@policychat.com with the page URL, the specific number, and the correct value. We publish corrections within 48 hours of verification.

7. Independence disclosure

PolicyChat is a comparison platform, not an insurance carrier. We earn revenue through licensed lead distribution to carriers via EverQuote and LendingTree. Our methodology is independent of which carriers compensate us — we do not adjust rate display order, ranking, or “best for X” routing based on partner commissions. Where partner rates appear, the partner is disclosed inline.

8. Versioning

This methodology document is versioned. Material changes are dated and disclosed in the version history. The current version is in effect from 2026-05-21.

Maintained by PolicyChat Editorial. Last reviewed: 2026-05-21. Editorial contacts: editorial@policychat.com.