The Score

How the number is made

Every Score is computed in three steps, in the same order on every path: the free Score, the paid report, the widget, the certificate and Geo.

1. Read the factors. Nine factors are read from public records and scored 0–100 each. A factor with no data is marked included: false and left out.
2. Weight the composite. Seven of these are weighted into the parcel score; seismic and radon are published at regional resolution and are reported alongside the score as context rather than averaged into it. The weighted mean of the included parcel factors, rounded, is the composite.
3. Apply the landmine caps. Some conditions make a lot hard to build no matter how good everything else is: a floodway, a wetland, a 30% slope. Each one caps the Score. The published Score is min(composite, every cap that fired).

The factors

Weights

Most states use the national weights. Florida, California and Texas carry their own, because flood, wildfire and utility reach move outcomes there more than elsewhere. The result names the weights that applied in region. Seismic and radon carry no weight anywhere: they are published at county or regional resolution, so they sit beside the number as a context band (low, elevated or high) instead of moving it.

Percentages are rounded to the nearest point.

Landmine caps

A cap is a ceiling a record puts on the Score. Blockers are conditions that usually stop a project. Cautions usually add cost, time or a study. The report lists every cap that fired with the record behind it, and the result keeps the composite as formula.uncappedScore, so you can always see what the lot scored before the cap.

The septic and dwelling “Very limited” caps do not fire on soil mapped as “Urban land”, where the survey rates the pavement rather than the ground. Some findings are disclosures that never cap: sewer service assumed, not confirmed, septic servicing unknown, private access, split zoning, road frontage not established, wetlands not screened. The cap rules carry their own version, bf-4.

Grades, bands and status

The same number is expressed three ways for three audiences. The thresholds are part of the published methodology and change only with a version.

Reading a result

The fields to read first. This is the shape of a report result; the certificate prints the same score, grade and band, and the widget shows the score and band.

{
  "result": {
    "score": 60,
    "grade": "C+",
    "band": "Good",
    "status": "caution",
    "version": "v3",
    "region": "TX",
    "factors": [
      {
        "key": "terrain",
        "score": 92,
        "included": true
      },
      {
        "key": "flood",
        "score": 56,
        "included": true
      },
      {
        "key": "radon",
        "score": 90,
        "included": true
      },
      {
        "key": "zoning",
        "score": 0,
        "included": false
      }
    ]
  },
  "formula": {
    "uncappedScore": 83,
    "capped": true,
    "boundBy": "sfha_majority"
  }
}

• Show result.score. When formula.capped is true, say which cap set it (formula.boundBy).
• Check included before showing any factor’s sub-score.
• Store result.version next to every score you keep.

Versions

The current methodology is v3. It splits the model in two levels: the parcel-scale factors carry the weight, and the two hazards published only at regional resolution sit beside the number as context. v2 added the lot-scale terrain and geometry factors, graded zoning from its dimensional standards, and dropped optimistic defaults. Earlier versions stay registered, so a stored factor set re-scores under the version it was made with.

The same methodology is published as data at /docs/score/methodology.json: factors, weights, caps and thresholds read from the engine at build time, so the file cannot drift from the number the product computes.

A stored parcel is never re-scored behind your back. When the methodology changes, a new version ships alongside, and a Score you have seen does not move.

What the Score is not

The Score is a screen built from public records. It is not an appraisal, a survey, an engineering opinion or an official zoning determination. It measures proximity to utilities, not capacity or a right to connect, and road adjacency, not legal access. The Accuracy page has the backtests, and Accuracy & limits lists what no screen here can tell you.