Methodology

Every dollar figure on Home Repair Index is the output of a documented, repeatable pipeline. Nothing is typed in by hand, and nothing is generated by guesswork. This page explains the pipeline end to end, so any number on the site can be traced back to its inputs.

Step 1 — National baselines from published sources

For each repair type (say, pier installation or flashing replacement) we cross-reference at least three independent published cost guides — sources such as HomeGuide, This Old House, Bob Vila, Today's Homeowner, and Angi. Each source's figure is recorded verbatim, with its URL and the date we retrieved it. From those figures we record a low, typical, and high national baseline.

When sources disagree by more than 3x, we widen the range to cover the disagreement and publish a note explaining why (for example, spot polyurethane lifts vs. full slab mudjacking are priced very differently by different guides). We would rather publish an honest wide range than a falsely precise number.

Step 2 — Local adjustment from government wage data

Labor is roughly half the cost of most repair jobs, so local labor costs move local prices. For each metro we pull the average weekly construction wage (NAICS 23, private employers) from the Bureau of Labor Statistics' Quarterly Census of Employment and Wages — public government data, updated quarterly.

Each metro's wage is compared to the national average, the deviation is damped by 0.5 (because materials and overhead don't vary as much as labor), and the result is clamped to a band of 0.75–1.35 to prevent any single data quirk from producing an absurd multiplier. The multiplier for every metro is printed on its pages.

Where BLS suppresses a metro's construction data for confidentiality, we fall back to the state-level wage, and say so on the page. Where the state figure is also suppressed (it happens — Missouri, for example), we use the national baseline unadjusted, and say that too.

Step 3 — Local context, not local fiction

Each metro page is written for that metro, around public data for its principal county (the county that holds the city's center, resolved through the Census geocoder):

• Soil. The USDA NRCS soil survey (SSURGO, via Soil Data Access) rates every mapped soil for building a house without a basement and lists what limits it. We report the share of the county's rated soil that is "very limited" by shrink-swell alone, and the most widespread of those soils by name. Shrink-swell clay is the dominant driver of foundation damage in much of Texas, Oklahoma, and Kansas. The rating covers the top 80 inches or so of soil across the whole county, rural land included, so it can understate problems that come from swelling bedrock below that depth (parts of Colorado's Front Range) and it does not see fill or drainage problems on a single lot.
• Drought. The U.S. Drought Monitor publishes a weekly map; for each county we count the weeks in the last ten years when at least half the county was in severe drought or worse (D2+). Clay shrinks as it dries, so long drought stretches are when clay-soil foundations move.
• Hail and wind. The NOAA NCEI Storm Events Database records reported hail (with stone size), damaging thunderstorm wind, and tornadoes by county. We count the days with hail 1 inch or larger over the last ten full years, the largest stone, and the most recent large-hail day. These are reports, so they reflect how many people were there to report a storm as well as the storm itself.
• Climate. NOAA's 1991-2020 U.S. Climate Normals from the nearest first-order weather station (or, where none is close, the nearest Cooperative Observer station) give heating and cooling degree days, hot days, and freezing nights: how hard a heating or cooling system works in a typical year. Each metro also carries its IECC climate zone and its U.S. Department of Energy region for air-conditioner efficiency standards (a SEER2 13.4 minimum in the North, 14.3 in the Southeast and Southwest, plus an EER2 floor in the Southwest), which sets the entry price of replacement equipment.

Every local number in a page's text is filled from these datasets at build time, the same way prices are, so a data refresh updates the words too. We do not invent neighborhood anecdotes, local ordinances, or storm stories.

Step 4 — Rounding and presentation

Figures under $1,000 are rounded to the nearest $50; figures above, to the nearest $100. We present ranges with a typical midpoint, never single-dollar precision — a $4,837 "average" is a false signal, and we won't print one.

Step 5 — Re-verification on a schedule

The entire pipeline re-runs quarterly: sources are re-retrieved, wage data is refreshed to the newest published BLS quarter, and every page's "Last verified" stamp reflects the run date. If a source changes its figures, ours change too.

What we are not

We are not contractors, and we don't produce quotes. Costs for a specific home depend on inspection findings no data pipeline can see. Treat our figures as calibration — what a fair range looks like in your metro — and get multiple local bids for anything structural.

Corrections

Spot a number that doesn't trace? Email [email protected] with the page URL. Verified corrections ship in the next build.