WhereToThrive

Methodology

How WhereToThrive scores cities, labels source quality, and explains rent, walkability, confidence, and freshness.

What the match score means

WhereToThrive starts with your questionnaire answers across budget, climate, work, community, daily life, family, and practical needs. Each city receives comparable scores for those categories, and the final list ranks cities by how closely they match the answers you supplied.

The score is a screening tool for relocation research. It is not a guarantee that a city will fit every neighborhood, household, job market, school, or housing need.

Source quality labels

Housing cost labels

The primary rent figure shown today is ACS median gross rent. It reflects rents paid by existing renters in the Census ACS sample, including some utility costs. It is not a current asking-rent feed, not a one-bedroom quote, and not a Zillow or apartment-listing index.

Where available, city pages and results also show HUD Fair Market Rent by bedroom count. HUD FMRs are annual gross-rent benchmarks for standard-quality units and housing-assistance programs. The 2-bedroom FMR is useful as a public current-market benchmark, but it is not the same measure as ACS median gross rent and can use HUD FMR areas that subdivide a CBSA.

Walkability label

Walkability is a commute-derived walkability estimate from ACS walking, biking, and transit commute shares, blended with Census Gazetteer CBSA population density where area coverage is available. Some metros fall back to commute-mode data only. It is not official Walk Score data and should be interpreted as a broad signal for transportation behavior, not block-by-block pedestrian quality.

Climate risk (FEMA National Risk Index)

Climate-risk profiles come from the FEMA National Risk Index (NRI), a public-domain county-level dataset. For each county and each of 18 natural hazards, FEMA publishes an Expected Annual Loss (EAL): the dollars of building, population, and agriculture losses statistically expected in an average year, combining hazard frequency, exposure, and historic loss ratios. NRI also publishes county Social Vulnerability and Community Resilience scores.

Aggregation. Counties are mapped to metros with the Census 2023 CBSA delineation file. For each metro, WhereToThrive sums county EAL dollars and divides by NRI county population to get expected annual loss per resident — overall and for ten relocation-relevant hazards (hurricane, coastal and inland flooding, tornado, wildfire, heat wave, earthquake, winter weather, drought, ice storm). Each per-resident rate is then percentile-ranked across the 250 screened metros (0-100, higher = more expected loss). The city-level climate-risk composite is the inverted all-hazard percentile, so higher means lower expected loss per resident. Social Vulnerability and Community Resilience are population-weighted county means. The legacy 1-3 natural-disaster band shown elsewhere is derived from the composite by tertile.

Why per-resident rates instead of FEMA's published scores. FEMA's own NRI score fields are national percentiles of absolute EAL dollars. Because EAL includes exposure, any large, dense, high-property-value metro can rank "high" on nearly every hazard from sheer size alone — Miami-Dade ranks in the top few percent nationally for absolute wildfire EAL even though FEMA rates its wildfire risk only "Relatively Moderate". Dividing by population removes most of that artifact, at the cost of no longer matching FEMA's published county percentiles exactly. Both views are honest; this site's tables label which one is shown.

Worst-pocket honesty. A population-weighted metro mean can hide a bad county (a metro can screen moderate for wildfire while one member county screens very high). Where a county's FEMA percentile for a hazard is materially above the metro mean and FEMA's own rating for that county is at least "Relatively Moderate", city pages show a worst-county callout using FEMA's unmodified county score and rating. Drought is excluded from the callout scan because its EAL is dominated by agricultural losses.

Limits. EAL is a statistical average of annualized losses, not a prediction, and it reflects historical patterns plus current exposure — it does not model forward-looking climate scenarios. Insurance availability and pricing move faster than any public dataset; treat the profile as a research screen and verify current insurance quotes for a specific address.

Coverage and source quality

Data coverage shows how many of the core source families are available for a city, shown as N of 11 (a completeness count, not a confidence score). Source quality labels show how individual fields are produced: direct, curated, formula, proxy, fallback, or unavailable. A city can have full coverage and still include formula/proxy scores such as walkability, food, nightlife, outdoor, school, or cost index. Food and nightlife now use Census County Business Patterns restaurant and drinking-place density where available; school scoring blends ACS adult educational attainment with NCES public-school access. Curated editorial overrides are labeled at the field level.

Refresh cadence

The public source manifest for this build was generated on 2026-07-13. Each source card lists the source year, collection date, quality label, and refresh cadence. Most government datasets refresh annually or on official release cycles; curated formulas and overrides refresh when the scoring model changes.

Monthly Cost Snapshot

The Monthly Cost Snapshot (on the calculator and in reports) assembles a household-specific monthly fixed-cost screen from labeled public data. It is deliberately not called a cost-of-living total: it covers housing, state income tax, sales tax, electricity, childcare, and commuting only. Every dollar figure is rounded to the nearest $25, and every row is tagged benchmark (a public figure used directly) or modeled (a documented formula with stated assumptions).

Line items and sources

Energy usage model (modeled)

The dataset stores two NOAA-normals-derived temperature markers per metro: typical July daily high and typical January daily low. Degree days and usage are estimated as:

CDD ≈ max(0, (summerHigh − 10) − 65) × 30 × 4.5  ·  HDD ≈ max(0, 65 − (winterLow + 10)) × 30 × 5.5
annual kWh ≈ 6,600 + 0.45 × CDD + 0.18 × HDD  ·  monthly $ = kWh/12 × state residential price

The ±10°F terms convert daily extremes to approximate monthly means; 30 approximates days in the peak month; 4.5/5.5 express the full cooling/heating season as multiples of the peak month; the heating coefficient is deliberately low because many homes heat with gas or oil. This is a screening estimate, not a utility quote.

Commute model (modeled)

monthly $ = cars × (avg one-way minutes × 2 × 30 mph ÷ 60) miles × 21 workdays × $0.725/mile (IRS 2026)

Why there is no groceries line

The familiar city grocery indexes (C2ER/COLI and similar) are licensed commercial datasets, and WhereToThrive only ships public, redistributable data. The BEA Regional Price Parities goods component is shown as context in its place. Numbeo-style crowdsourced prices are excluded for quality reasons.

Monthly Cost Snapshot limitations

Source manifest