Buildability- TerrainUSGS 3DEP20%
- ZoningCity & county code16%
- FloodFEMA16%
- SoilUSDA12%
- WildfireUSFS12%
- UtilitiesOpenStreetMap12%
- Lot shapeParcel record11%
- SeismicUSGSContext
- RadonEPAContext
The sample report · 1247 Pine St, an invented parcel
The Buildability Score™
One number. Published math.
A 0–100 read on how buildable a U.S. parcel is, from public records you can check. Scroll to watch the dial build it.
Free · No card, no signup · About 20 seconds
Nine factors are read.
For each address the engine asks the federal and local records: flood, terrain, soil, wildfire, seismic, radon, utilities, zoning and the lot itself.
Seven carry weight.
By national default, terrain counts most, at 20%. Seismic and radon are read and shown as regional context, but not averaged in.
A missing record drops out.
If a source has nothing for the parcel, that factor leaves the average and the rest share its weight. It never counts as a zero, and never as a pass.
A landmine caps it.
Some findings end the conversation. A regulatory floodway caps the Score at 25, however good the rest looks.
Try a landmine.
Some findings end the conversation. Switch one on: the sample lot’s 87 drops to its cap.
Weighted average 87. No caps. Strong
80+ Strong · 60+ Viable · 40+ Moderate · <40 Constrained
A screen from public records, not a permit or appraisal · Austin backtest: built lots ranked above undeveloped ones, AUC 0.711 · How we test it
Go deeper
The Score, in full
The definition, the scale, every factor and weight, the caps and the tests behind them, as the engine publishes them.
What the Score is
The Buildability Score™ is a 0–100 rating of development feasibility for any U.S. parcel, computed from nine factors — zoning, flood, seismic, soil, radon, wildfire, utilities, terrain, and lot geometry — sourced live from seven federal agencies plus parcel and zoning partners, with every finding cited to its source.
Higher scores mean easier feasibility; lower scores flag hazard, regulatory, or site constraints. It covers all 50 states, and takes about 20 seconds from address to verdict. The seven agencies, and what each supplies:
- Census
- Geocoding · TIGER boundaries · block groups
- FEMA
- NFHL flood zones · SFHA designations
- USGS
- Seismic hazard · 3DEP terrain and slope
- EPA
- Radon zones · community water systems
- USDA
- SSURGO soils · suitability ratings
- USFS
- Wildfire Hazard Potential
- USFWS
- National Wetlands Inventory
Buildability reads public records. Records can be wrong or out of date. Verify before you act. The Score is a screening standard, not a replacement for entitlement counsel on the deal you pursue.
0 to 100, end to end
Every score maps to a letter grade, a band and a plain answer. The plain answer turns at 70 and 50; the bands at 80, 60 and 40.
| Score | Grade | Band | Plain answer |
|---|---|---|---|
| 90–100 | A+ | Strong | Likely yes |
| 85–89 | A | Strong | Likely yes |
| 80–84 | A- | Strong | Likely yes |
| 75–79 | B+ | Viable | Likely yes |
| 70–74 | B | Viable | Likely yes |
| 65–69 | B- | Viable | Maybe |
| 60–64 | C+ | Viable | Maybe |
| 55–59 | C | Moderate | Maybe |
| 50–54 | C- | Moderate | Maybe |
| 40–49 | D | Moderate | Hard to build here |
| 30–39 | D | Constrained | Hard to build here |
| 0–29 | F | Constrained | Hard to build here |
Thresholds are part of the published methodology; changing one requires a version bump.
SourcePublished thresholds, methodology v3·the engine’s grade and verdict labels
Nine factors, and what each reads
Each factor measures a different side of feasibility. Their weights, nationally and by region, are further down.
| Factor | Source | What it measures |
|---|---|---|
| Terrain | USGS 3DEP, queried live | Ground slope at the lot. Slope differs house to house, so it separates neighbors. |
| Zoning | County and municipal records via partners | Dimensional standards on record: FAR, lot coverage, minimum lot size. Omitted when only a designation name is known. |
| Flood | FEMA NFHL, queried live | Flood zone, Special Flood Hazard Area status, and the share of the lot inside an SFHA. |
| Soil | USDA SSURGO, queried live | Soil survey classification, drainage, and building suitability. |
| Wildfire | Federal wildfire hazard layers | Wildfire hazard potential around the parcel. |
| Utilities | OpenStreetMap infrastructure | Proximity to power, water, and road access. |
| Lot shape | Parcel ring via partners | Lot shape regularity. A long sliver builds differently than a clean rectangle. |
| Seismic · context | USGS, queried live | Earthquake hazard level at the site. |
| Radon · context | EPA radon zones, queried live | County radon zone designation. |
How a reading becomes a sub-score
Each factor is scored 0–100 before it is weighted. Severe findings pull their factor deep down instead of averaging away, and a factor with no record drops out rather than being guessed.
- Flood
- By share of the lot in an SFHA: 95 − 65 × the share, so 0% reads 95 and the whole lot reads 30. Without a lot share: SFHA at the centroid 40, zone A or V 30, zone X 95 · any other mapped zone 90.
- Wildfire
- Low 85 · moderate 70 · high 50 · very high 35. Unknown drops out.
- Terrain
- Slope 0% → 96 · 4% → 92 · 8% → 84 · 15% → 65 · 25% → 42 · 40% → 20 · 60% → 10, straight lines between.
- Lot shape
- Regularity 0.2 → 40 · 0.4 → 55 · 0.6 → 72 · 0.8 → 88 · 1.0 → 95.
- Zoning
- The mean of the FAR, lot coverage and minimum lot readings on record. A larger minimum lot reads lower.
- Soil · Utilities
- Their sub-scores pass straight through: SSURGO building suitability, and utility readiness.
- Seismic · context
- Low 90 · moderate 75 · high 55 · very high 40.
- Radon · context
- Low 90 · moderate 75 · high 60.
Heavy hits and light lifts
The same sub-scores, in plain words: which findings sink a factor, which only dent it, and which lift it.
- Heavy penalties · sub-score in the 30s–40s
- Some conditions dominate their factor: a FEMA Special Flood Hazard Area (A/V zones), very high wildfire hazard potential, a poor soil suitability rating, no utility access nearby. With regional weighting a high-hazard flood zone can dominate the verdict. The hardest constraints skip weighting entirely: a regulatory floodway or a mapped wetland caps the score outright.
- Moderate penalties · sub-score in the 50s–70s
- Moderate wildfire exposure, a middling soil drainage class, steep ground slope at the lot. These reduce feasibility without dominating it, and warrant more diligence, mitigation or budget.
- Strong signals · sub-score in the 85–95 range
- FEMA zone X with minimal flood risk, low wildfire risk, a high soil suitability rating, utilities confirmed close by. Low-risk findings lift the score.
SourceScore methodology
Weights by region
Flood carries 25% of the score in Florida. Utilities carry 24% in Texas. Terrain and zoning rise in California.
Seismic and radon are read, scored and cited, but carry no weight. Radon is the EPA county zone, identical for every lot in a county; seismic is a USGS design category set at regional, not lot, resolution. Neither can rank two neighbors, so both are reported beside the score as context instead of averaged into it.
| Factor | National | Florida | California | Texas |
|---|---|---|---|---|
| Terrain | 20% | 11% | 25% | 15% |
| Zoning | 16% | 14% | 22% | 14% |
| Flood | 16% | 25% | 10% | 14% |
| Soil | 12% | 14% | 9% | 12% |
| Wildfire | 12% | 13% | 20% | 12% |
| Utilities | 12% | 12% | 5% | 24% |
| Lot shape | 11% | 13% | 10% | 10% |
| Seismic | Context | Context | Context | Context |
| Radon | Context | Context | Context | Context |
| Total, unrounded | 1.0 | 1.0 | 1.0 | 1.0 |
Shown rounded to whole percent, so a column can add to 99–102%. The engine uses the exact weights, which sum to 1.0.
SourceScore methodology, default and regional weights·Machine-readable methodology
The same lot, four regions
One parcel with strong land signals and weak utilities, scored by the engine under each weight set. Same parcel, same sub-scores; only the weights change.
| Factor | Reading | Sub-score | National | Florida | California | Texas |
|---|---|---|---|---|---|---|
| Terrain | 3% slope | 93 | 20% | 11% | 25% | 15% |
| Zoning | 40% coverage, 20,000 sf min lot | 68 | 16% | 14% | 22% | 14% |
| Flood | No part of the lot in an SFHA | 95 | 16% | 25% | 10% | 14% |
| Soil | Sub-score 80 | 80 | 12% | 14% | 9% | 12% |
| Wildfire | moderate | 70 | 12% | 13% | 20% | 12% |
| Utilities | Sub-score 25 | 25 | 12% | 12% | 5% | 24% |
| Lot shape | 0.85 regularity | 90 | 11% | 13% | 10% | 10% |
| Seismic | low | 90 | Context | Context | Context | Context |
| Radon | moderate | 75 | Context | Context | Context | Context |
| Score | 76 · B+ | 77 · B+ | 78 · B+ | 69 · B- |
In words: National Viable, “Likely yes” · Florida Viable, “Likely yes” · California Viable, “Likely yes” · Texas Viable, “Maybe”. Weights are shown rounded to whole percent, so a column can add to 99–102%; the engine uses the exact weights, which sum to 1.0.
SourceThe score explorer’s example parcel·computed by the v3 engine
The formula, on one line
Score = min( Σ wi·xi / Σ wi over factors with data, every landmine cap )
- A cap only pulls down. A landmine can lower the score to its cap, never raise it.
- No record never raises a landmine. A cap fires only on an affirmative signal from a named source.
- Every landmine is named. It shows as its own finding, with the source that raised it and what to do next.
Every landmine cap
Blockers cap at 45 or below. Cautions cap between 60 and 75; the no-community-water cap sits at 75 so it binds only on an otherwise excellent lot. When more than one fires, the lowest cap wins.
| Finding | Caps at | Class | Source |
|---|---|---|---|
| Regulatory floodway | 25 | Blocker | FEMA NFHL |
| Mapped wetland on the parcel | 40 | Blocker | USFWS NWI |
| Housing not permitted by right in the district | 40 | Blocker | National Zoning Atlas |
| Severe slope, 30% or more | 45 | Blocker | USGS 3DEP |
| More than half the lot in an SFHA | 60 | Caution | FEMA NFHL, parcel ring |
| Soil “Very limited” for septic, lot off the grid | 60 | Caution | USDA SSURGO |
| Soil unit floods frequently | 60 | Caution | USDA SSURGO |
| Soil “Very limited” for dwellings | 65 | Caution | USDA SSURGO |
| Bedrock within 50 cm | 70 | Caution | USDA SSURGO |
| Seasonal water table within 60 cm | 70 | Caution | USDA SSURGO |
| No community water system | 75 | Caution | EPA |
Six disclosures never cap: wetlands not screened, sewer assumed, septic limitation not applied when servicing is unknown, split zoning, road frontage not established, private access. They name a gap so an assumption is not invisible.
What the engine guarantees
- Versioned and frozen. The current engine is v3. The same inputs produce the same number; a change to the math needs a reviewed version bump, never a silent tweak.
- Golden-tested in CI. Hand-computed golden cases guard the engine. If a code change moves any expected score, the build fails.
- Missing data drops out. A source with no data leaves the score; a regional-estimate provenance is labeled Estimated. Nothing is silently penalized.
- Every finding cited. Each factor names the dataset it came from: FEMA NFHL, USGS, EPA, USDA SSURGO. Sources are queried live when a report is generated; successful lookups are cached server-side for up to 7 days, so repeat lookups are instant.
- Deterministic and pure. No network, no clock, no randomness inside the engine. The number maps to a grade, band and verdict through published thresholds.
SourceScore methodology·the engine’s golden test suite
Versions and the backtest
Methodology v3 is current, with formula bf-4. v1 and v2 stay frozen and addressable, so a historical score reproduces exactly. v2 (flipped 2026-07-20) weighted all nine factors; v3 reads nine and weights seven. v1 assumed 85 on missing utilities; v2 and v3 drop the factor instead.
Discrimination is unchanged between v2 and v3; the intervals overlap. What v3 improves is spread. One metro and 76 positives, so this is evidence, not proof.
| Version | Factors weighted | AUC | 95% interval | Spread (σ) |
|---|---|---|---|---|
| v3 · current | 7 of 9 | 0.711 | 0.650–0.766 | 6.62 |
| v2 · frozen | 9 of 9 | 0.706 | 0.644–0.762 | 5.58 |
| v1 · frozen · original live test | 7 of 7 | 0.741 | 0.67–0.81 | — |
All three rows are the same 76 built vs 369 undeveloped single-family parcels in Austin, TX. v1 is the original live test, run in July 2026; v2 and v3 are re-scored offline from the stored per-factor sub-scores. v1’s cohort-wide spread was not published, so its σ is left blank.
Every number, and the record behind it
A number without a record does not stay put. Each of these is derived from the code that ships, and a contract test fails the build if a claim drifts.
| Number | What it counts | Where it comes from | What it is not |
|---|---|---|---|
| 9 | Factors read, scored and cited on every report | The shipping methodology’s factor list | Not all of them move the number; see the next two rows |
| 7 | Factors weighted into the parcel score | The methodology’s weight map, which sums to 1.0 | Not the same as the count above. The previous methodology weighted every factor it read; this one does not |
| 2 | Factors carried as regional context, at weight zero | Published at region resolution: a county radon zone and a four-step national seismic class | Not scoring factors. They cannot rank two neighbors |
| 31 | Datasets the report queries for a parcel | A registry pinned to the report’s own provider list by a contract test: add a provider without registering it and the build fails | Not a count of datasets that returned data. Coverage varies by parcel, and a dataset can answer, answer empty, or not cover the location |
The last column is the one that matters. Earlier versions of this site claimed four different factor counts and two different source counts, because each was written to sound right rather than derived from the thing it described. Retiring them cost nothing real: the work they were pointing at is still here, and now it can be checked.
Questions
The Score
What is the Buildability Score™?
A 0–100 feasibility score for any U.S. parcel: nine factors rolled into one number with a grade and a plain answer. 70+ reads “Likely yes”, 50–69 reads “Maybe”, below 50 reads “Hard to build here”.
How is it calculated?
Nine factors are read and seven are weighted; seismic and radon are regional context. The weighted average runs over the factors with data, then any landmine caps it. Defaults: terrain 20%; zoning and flood 16%; soil, wildfire and utilities 12%; lot geometry 11%. Regional overrides apply (Florida flood at 25%, Texas utilities at 24%), and missing data drops out entirely.
Can the score change without notice?
No. Each methodology version (current: v3) is frozen and golden-tested: reference cases run in CI, and any change that moves a score fails the build. Improvements ship as a new version; historical scores reproduce exactly.
What does a high score mean?
A score of 70+ (grade B or better) reads “Likely yes”: strong development potential with few major obstacles. 50–69 reads “Maybe”: some factors need attention. Below 50 reads “Hard to build here”: significant challenges worth understanding before you spend on consultants.
Using it
How accurate is it?
Every factor comes from live federal queries (Census, FEMA, USGS, EPA, USDA, USFS, USFWS) plus parcel and zoning partners, and every finding cites its dataset. It is a screening tool, not a permit guarantee: verify locally before committing capital.
In the Austin backtest, built lots ranked above undeveloped ones at AUC 0.711. How we test it.
How long does it take?
About 20 seconds: geocode via Census, query the sources live, run the engine, return the score with its factor breakdown. The live Score is free and unlimited; the $99 report adds the parcel record.
Pricing
Is the Score free?
Yes. The live Score is free and unlimited on every address. The $99 Property Report adds the parcel record, at $79 each at volume, and reports never expire.
No record, no charge: if the parcel record cannot be found, you are not billed.
Keep readingAudit the data sources·How accurate it is·The worked example·Glossary·Sample report
Score a parcel.
Every finding cited to its source.