Document submitted
A homeowner provides an actual PDF, scan or photograph—not a remembered total.
OPEN METHODOLOGY
Where the numbers come from.
Price transparency only works if the underlying evidence is transparent too. This page explains what we collect, how comparisons are made, and what our confidence percentages do—and do not—mean.
THE PIPELINE
From proposal to public benchmark
Private details removed
Name, exact address, phone, email, signatures, account numbers and access instructions are redacted.
Scope and property extracted
We capture quantities, materials, exclusions, quote date, ZIP, building type, construction era, historic status and site constraints. The homeowner confirms the extraction.
Local comparable assigned
A record moves from metro to neighborhood only when the smaller group remains useful and private. Historic and new-build work are never silently blended.
Outcome followed
When possible, the homeowner later confirms whether the quote was accepted and what the completed project actually cost.
TWO DIFFERENT PERCENTAGES
Confidence is not accuracy.
We report them separately because combining them would make a young dataset look more certain than it is.
72%
Prototype confidence
A record-quality score using document verification, scope completeness, geographic match, recency, sample size and outcome verification.
Available now · illustrative until real records are verifiedN/A
Measured price accuracy
How often our predicted range contains the eventual contracted or completed cost. This requires real outcomes and a held-out test group.
Published only after at least 30 comparable completed projectsCONFIDENCE FORMULA
What raises or lowers the score
SignalWeightWhat earns full credit
Document verification20%Original proposal supplied and readable
Scope completeness25%Quantities, material grade, labor and exclusions captured
Geographic match15%Same neighborhood or a validated comparable local market
Property match15%Same building form, era and historic constraints
Recency10%Issued within the selected year range
Comparable sample size10%At least 10 closely matched records
Outcome verification5%Contracted or completed total confirmed
Neighborhood results require enough records to avoid exposing a household and to prevent one unusual proposal from dominating the benchmark. Otherwise, the site automatically falls back to the ZIP, city or metro level.
TIME ADJUSTMENTS
Year and season both matter.
We retain the original nominal price and quote date. We never overwrite history; adjusted figures are displayed alongside original figures.
Year ranges
Users can limit comparisons to a single year or a date range. Older records are inflation-adjusted with construction-specific indices where available, not the general consumer price index alone.
Selected range2022–2026
Time of year
Seasonality is calculated separately by trade and region. Exterior masonry may rise during high-demand months; interior work may follow a different pattern.
Market shocks
Material spikes, permit changes and unusual labor disruptions are flagged rather than quietly averaged into an ordinary seasonal pattern.
! Exceptional period labeled
FUTURE ACCURACY REPORT
What we will publish
Once enough completed projects exist, this page will show accuracy by category, geography and year—not one flattering site-wide number.
01Median absolute error in dollars and percent
02Share of outcomes within ±10%, ±15% and ±20%
03Sample size and date range for every result
04Performance on unseen, held-out projects
05Corrections and methodology changes over time