# How Insights scores are calculated

> The formulas behind the Composite Score and five category scores: which metrics feed each score, their weights, and how ranks and percentiles work.

## Before you start

- Scores exist at **State**, **Metro**, and **County** level only. **City**
  level lists indicators directly and has no scores, benchmarks, or
  strengths-and-weaknesses summary.
- Every score runs 0 to 100 and higher is always better, including for
  **Risk Score** — a high Risk Score means _less_ exposure, not more.
- A score depends only on that location's own numbers. Loading more
  locations, filtering, or comparing never changes anyone's score.

## How the scores are built

Each location gets five **category scores** — **Cost Score**, **Workforce
Score**, **Risk Score**, **Quality of Life Score**, and **Regulatory
Score** — plus a **Composite Score** that blends the five.

A category score is a weighted average of that category's metrics, after
each metric is rescaled onto a 0-to-100 scale. The Composite Score then
blends the five category scores, and the blend is **not** the same at every
level:

| Level  | Cost | Workforce | Risk | Quality of Life | Regulatory |
| ------ | ---- | --------- | ---- | --------------- | ---------- |
| State  | 20%  | 20%       | 20%  | 20%             | 20%        |
| Metro  | 20%  | 25%       | 15%  | 20%             | 20%        |
| County | 20%  | 25%       | 15%  | 20%             | 20%        |

So at Metro and County level, Workforce counts for more than Risk. At State
level the five are equal.

Two things worth knowing before you read the tables below:

- The metrics that _feed_ a score are not the same as the metrics shown
  under that heading in the detail panel. **Median Household Income** feeds
  the State **Quality of Life Score** even though it is listed under Site
  Screening, and **Workers' Comp Rate** feeds both the Cost Score and the
  Regulatory Score at Metro and County level.
- Some metrics are shown but score nothing. **Population**, **Median Age**,
  **Poverty Rate**, **Freight Volume Index**, **Per Capita Income**, **Metro
  GDP**, and **Total Employment** at county level are reference figures, not
  score inputs.

## State: which metrics feed each score

| Category        | Metric                      | Weight | Scale range     | Direction     |
| --------------- | --------------------------- | ------ | --------------- | ------------- |
| Cost            | Electricity Rate            | 25%    | 8–33 ¢/kWh      | Lower better  |
| Cost            | Natural Gas Price           | 15%    | $9–$50 /MCF     | Lower better  |
| Cost            | Tax Climate Rank            | 25%    | #1–#50          | Lower better  |
| Cost            | Workers' Comp Rate          | 15%    | $1–$3 per $100  | Lower better  |
| Cost            | Health Premium (Annual)     | 20%    | $6,500–$9,000   | Lower better  |
| Workforce       | Unemployment Rate           | 20%    | 1.5%–7%         | Lower better  |
| Workforce       | Labor Force Participation   | 25%    | 50%–72%         | Higher better |
| Workforce       | Bachelor's Degree+          | 25%    | 18%–45%         | Higher better |
| Workforce       | HS Graduation Rate          | 15%    | 70%–95%         | Higher better |
| Workforce       | Avg Commute Time            | 15%    | 15–35 min       | Lower better  |
| Risk            | Disaster Declarations (5yr) | 35%    | 2–30            | Lower better  |
| Risk            | Avg Air Quality Index       | 35%    | 25–65           | Lower better  |
| Risk            | Health Ranking              | 30%    | #1–#50          | Lower better  |
| Quality of Life | Broadband Access            | 20%    | 70%–92%         | Higher better |
| Quality of Life | Physicians per 10k          | 20%    | 15–45           | Higher better |
| Quality of Life | Violent Crime Rate          | 25%    | 80–850 per 100k | Lower better  |
| Quality of Life | Housing Affordability       | 20%    | 2–10            | Lower better  |
| Quality of Life | Median Household Income     | 15%    | $40,000–$90,000 | Higher better |
| Regulatory      | Economic Freedom Score      | 35%    | 3–9 out of 10   | Higher better |
| Regulatory      | Right-to-Work State         | 25%    | Fixed 70 / 30   | Yes is better |
| Regulatory      | Regulatory Burden Rank      | 40%    | #1–#50          | Lower better  |

## Metro: which metrics feed each score

| Category        | Metric                      | Weight | Scale range         | Direction     |
| --------------- | --------------------------- | ------ | ------------------- | ------------- |
| Cost            | Elec Rate (¢/kWh)           | 25%    | 8–30 ¢/kWh          | Lower better  |
| Cost            | Median Rent (2BR)           | 25%    | $700–$2,500         | Lower better  |
| Cost            | Median Home Value           | 20%    | $150,000–$1,100,000 | Lower better  |
| Cost            | Workers' Comp Rate          | 15%    | $1–$3 per $100      | Lower better  |
| Cost            | Tax Climate Rank            | 15%    | #1–#50              | Lower better  |
| Workforce       | Unemployment Rate           | 20%    | 1.5%–7%             | Lower better  |
| Workforce       | Labor Force Participation   | 25%    | 50%–75%             | Higher better |
| Workforce       | Bachelor's Degree+          | 25%    | 18%–55%             | Higher better |
| Workforce       | Median Annual Wage          | 15%    | $38,000–$72,000     | Higher better |
| Workforce       | Total Employment            | 15%    | 500,000–10,000,000  | Higher better |
| Risk            | Disaster Declarations (5yr) | 30%    | 0–30                | Lower better  |
| Risk            | Avg Air Quality Index       | 25%    | 20–80               | Lower better  |
| Risk            | County Health Ranking       | 20%    | #1–#25              | Lower better  |
| Risk            | Violent Crime Rate          | 25%    | 100–650 per 100k    | Lower better  |
| Quality of Life | Physicians per 10k          | 25%    | 12–45               | Higher better |
| Quality of Life | Broadband Access            | 25%    | 70%–95%             | Higher better |
| Quality of Life | Home Value / Income         | 25%    | 2–11                | Lower better  |
| Quality of Life | Colleges & Universities     | 25%    | 5–120               | Higher better |
| Regulatory      | Tax Climate Rank            | 30%    | #1–#50              | Lower better  |
| Regulatory      | Economic Freedom            | 30%    | 3–9 out of 10       | Higher better |
| Regulatory      | Right-to-Work               | 20%    | Fixed 70 / 30       | Yes is better |
| Regulatory      | Workers' Comp Rate          | 20%    | $1–$3 per $100      | Lower better  |

**Avg Commute Time**, **Per Capita Income**, and **Metro GDP ($M)** are
shown in the Metro panel but feed no score.

The four Regulatory metrics come from the metro's state, not from the metro
itself — the tab is labelled **Regulatory (State)** for exactly that reason.

## County: which metrics feed each score

| Category        | Metric               | Weight | Scale range         | Direction     |
| --------------- | -------------------- | ------ | ------------------- | ------------- |
| Cost            | Electricity Rate     | 20%    | 7–30 ¢/kWh          | Lower better  |
| Cost            | Median Rent (2BR)    | 25%    | $600–$3,000         | Lower better  |
| Cost            | Median Home Value    | 20%    | $100,000–$1,100,000 | Lower better  |
| Cost            | Cost of Living Index | 20%    | 82–165              | Lower better  |
| Cost            | Workers' Comp        | 15%    | $1–$3 per $100      | Lower better  |
| Workforce       | Unemployment         | 18%    | 1.5%–9%             | Lower better  |
| Workforce       | Labor Force Part.    | 22%    | 45%–75%             | Higher better |
| Workforce       | Bachelor's+          | 25%    | 12%–60%             | Higher better |
| Workforce       | Median Wage          | 17%    | $32,000–$80,000     | Higher better |
| Workforce       | Pop. Growth (5yr)    | 18%    | −5% to 12%          | Higher better |
| Risk            | Disasters (5yr)      | 30%    | 0–30                | Lower better  |
| Risk            | Air Quality Index    | 25%    | 20–80               | Lower better  |
| Risk            | Health Percentile    | 20%    | 2–99                | Higher better |
| Risk            | Violent Crime        | 25%    | 50–900 per 100k     | Lower better  |
| Quality of Life | Physicians /10k      | 25%    | 6–50                | Higher better |
| Quality of Life | Broadband Access     | 25%    | 60%–98%             | Higher better |
| Quality of Life | Home/Income Ratio    | 28%    | 2–10                | Lower better  |
| Quality of Life | Avg Commute          | 22%    | 14–40 min           | Lower better  |
| Regulatory      | Tax Climate Rank     | 30%    | #1–#50              | Lower better  |
| Regulatory      | Economic Freedom     | 30%    | 3–9 out of 10       | Higher better |
| Regulatory      | Right-to-Work        | 22%    | Fixed 72 / 32       | Yes is better |
| Regulatory      | Workers' Comp        | 18%    | $1–$3 per $100      | Lower better  |

**Population**, **Total Employment**, and **Per Capita Income** are shown
but feed no score.

Note that **Health Percentile** at County level runs the opposite way from
**Health Ranking** at State and Metro level: at County level a higher number
is healthier, at State and Metro level #1 is healthiest. See
[County and metro data quality](https://kb.incenti.co/insights/county-and-metro-data-quality.md).

## Turning a raw number into a 0–100 score

Every scored metric goes through the same four steps.

### Clamp to the scale range

A value below the bottom of the range is treated as the bottom; a value
above the top is treated as the top. This is why two locations can both show
**0** on the same metric: if the range for a State's Electricity Rate is
8–33 ¢/kWh, anything at or above 33 ¢ scores 0, whether it is 33 ¢ or 41 ¢.
The same applies at the top end — both score 100.

### Rescale to 0–100 and invert where lower is better

The clamped value becomes its position within the range, expressed as 0 to
100\. For a "lower is better" metric the result is flipped, so the cheapest
electricity or the fewest disaster declarations scores 100.

### Substitute a fixed value for Right-to-Work

**Right-to-Work** is a yes/no fact, not a measurement, so it contributes a
constant rather than a rescaled number: 70 for yes and 30 for no at State
and Metro level, 72 and 32 at County level.

### Average by weight, then round

The rescaled values are averaged using the weights in the tables above.
State scores are rounded to one decimal place (for example 72.4); Metro and
County scores are rounded to whole numbers. Comparison views round
everything to whole numbers for display.

> **Note:** Two Metro metrics — **County Health Ranking** and **Physicians per 10k** —
> are only available for 34 large metros. When they are missing, the Risk and
> Quality of Life scores are averaged over the metrics that _are_ present and
> the missing metric's weight is spread across them, rather than counting the
> gap as a zero. A metro with those metrics and one without are therefore not
> computed from an identical set of inputs.

## Score colours, ranks, and percentiles

### Badge colours

Every score badge is coloured by the number alone, at every level:

| Colour | Score  | Reads as      |
| ------ | ------ | ------------- |
| Green  | 70–100 | Strong        |
| Blue   | 50–69  | Above average |
| Amber  | 35–49  | Mid-range     |
| Red    | 0–34   | Below average |

A red badge means the score is under 35 on that dimension. It is not an
error and it is not a data problem.

### State rank and percentile

The State **Benchmarks** section reads **Ranked #X of N states**, where N is
the number of states currently being served, and works out each row's rank
by comparing that state against every other served state on the same metric
(1 is always best, whichever direction the metric runs). The **Nth pct**
pill next to a rank is `(N − rank) ÷ (N − 1)`, expressed as a percentage, so
the best-ranked state is at the 100th percentile and the worst is at 0. The
**nat'l** figure on each row is the simple average of the served states.

### County percentile

County **Benchmarks** show a percentile only — no rank. The percentile is
the share of all U.S. counties in the current universe with a lower raw
value on that metric, flipped for "lower is better" metrics so that a high
percentile always reads as good. The header states the pool size, for
example **Percentile rank against 3,143 U.S. counties**.

### EDD percentile

For Economic Development Districts, the percentile compares a district's
_share_ of coded goals in a category against every other **coded** district
— districts with no coded goals are not in the pool. See
[Explore EDD regions](https://kb.incenti.co/insights/edd-regions.md).

## Why the same place scores differently at each level

A state's score and the score of a metro or county inside it are different
numbers produced by different formulas, so they are not meant to match:

- The composite weights differ (equal fifths at State, Workforce 25% and
  Risk 15% at Metro and County).
- The scale ranges differ. Electricity is scaled 8–33 ¢/kWh at State, 8–30
  at Metro, and 7–30 at County, so the same rate lands on a different 0–100
  value at each level.
- The metrics differ. State Cost Score uses natural gas and health premiums;
  Metro and County Cost Scores use rent, home value, and either tax climate
  or cost of living instead.
- **Right-to-Work** contributes 70/30 at State and Metro but 72/32 at
  County.

Compare a location against others _at the same level_. Comparing a state's
72.4 with a metro's 68 is not a like-for-like comparison.

## Where the written call-outs come from

The **Insights** cards, the **Benchmarks** rows, and the **SWAT Analysis**
section are all generated by fixed rules from the numbers on screen. Nothing
in these sections is written by an AI assistant, and the same location
always produces the same wording from the same data. (The **AI Analysis**
section on a City panel is likewise rule-based.)

**SWAT Analysis** — the heading in the product — is a standard SWOT:
strengths, weaknesses, opportunities, and threats. Its rules are:

- **Strength**: any category score of 65 or more. Population growth above
  5%, broadband access of 90% or more, or unemployment at or below 3.5% each
  add one.
- **Weakness**: any category score of 35 or less. Population decline beyond
  −1% or broadband access under 70% each add one.
- **Threat**: 15 or more FEMA disaster declarations in five years, a violent
  crime rate above 500 per 100k, or an electricity rate above 12 ¢/kWh.
- **Opportunity**: combinations of scores, for example a Cost Score of 60+
  alongside a Workforce Score of 55+.

Empty quadrants read **None identified.** The **Strategic Trade-off** card
above the quadrants compares the Cost Score with the Workforce Score and
calls the location balanced when the two are within 8 points.

## Common questions

**Q: Is Workforce weighted more heavily than Risk in the Composite Score?**
A: At Metro and County level, yes — Workforce is 25% and Risk is 15%, with
Cost, Quality of Life, and Regulatory at 20% each. At State level all five
categories are weighted equally at 20%.

**Q: Why does my state score 72.4 at state level but the metro inside it
scores 68?**
A: They are different calculations. Each level has its own metric list, its
own scaling ranges, and — for Metro and County — unequal composite weights.
The two numbers are not comparable, and neither one is wrong. Compare
locations only against others at the same level.

**Q: Why is my score red?**
A: Score badges are coloured by value alone: green at 70 and above, blue at
50–69, amber at 35–49, and red below 35. A red badge means that dimension
scored under 35 for that location; it is not a data error.

**Q: Does a higher Risk Score mean more risk or less?**
A: Less. Every score in Insights runs so that higher is better, including
Risk Score. A high Risk Score means fewer disaster declarations, better air
quality, and a better health ranking.

**Q: Two locations show a score of 0 on the same metric even though their
raw numbers are very different. Why?**
A: Each metric is scaled against a fixed range, and values past the end of
the range are treated as the end of the range. Anything at or beyond the
worst end of the scale lands on 0, so two locations that are both well past
that point score identically. The raw numbers in the detail panel still show
the real difference.

**Q: Can I change the weights to match my project?**
A: Not in Explore — the scores there always use the weights above. The
**Analyze** area applies its own project-type weighting when it works out
which locations you compete with. See
[Analyze your competitive set](https://kb.incenti.co/insights/analyze-competitors.md).

**Q: Are the Insights cards and the SWAT Analysis written by AI?**
A: No. Every call-out in the **Insights**, **Benchmarks**, and **SWAT
Analysis** sections is produced by fixed rules from the numbers already on
the page, so the same data always produces the same wording.

## Related

- [Insights overview and data coverage](https://kb.incenti.co/insights/overview-and-data-coverage.md)
- [Where Insights data comes from and how often it refreshes](https://kb.incenti.co/insights/data-sources-and-refresh.md)
- [County and metro data quality](https://kb.incenti.co/insights/county-and-metro-data-quality.md)
- [Explore a location](https://kb.incenti.co/insights/explore-a-location.md)
- [Compare locations](https://kb.incenti.co/insights/compare-locations.md)
