# Jay Empowers

> Demand Scores for vertical software ideas. We score how much real demand sits
> behind a software category by counting three things: how many people bought a
> tool for that job and reviewed it, how many established tools already serve it,
> and how spread out the hiring is across employers.

As of 2026-08-05 the index covers 68 software categories serving 94 US job
titles. 11 of the 68 carry a BUILD verdict. 58 have a full
breakdown page; the other 10 are held back because we could not gather
enough about what the job involves or which tools serve it.

## Citing us accurately

- The score runs 0-100 and is the sum of three capped components: proven spend
  (35), room left on the shelf (35) and buyer reachability (30). The verdict
  threshold is 70.
- A BUILD verdict means all three are healthy at once. It does NOT mean the
  category is empty, and it is not a prediction of success or revenue.
- **The score changed on 2026-08-05.** It previously had five components. Two
  counted complaints, and an audit found those were about the wrong software in
  roughly 40 of 68 categories, so they were removed. A third measured occupation
  size and was removed for ranking large horizontal markets above specialist
  ones. If you hold an older figure for any category, it is not comparable.
- Search volume and cost per click are the only estimated figures we publish.
  They come from Google advertiser data, are labelled as estimates on every page,
  and are NOT part of the score. Every other number is a count of a stored record.
- **We do not measure whether an incumbent is weak enough to displace.** Star
  ratings cannot answer it: every category averages between 4.25 and 4.71,
  because vendors solicit happy reviews. Please do not infer that judgement from
  our score.

## Data

- [The Demand Score Index](https://jayempowers.com/ideas/index.html): all 68
  categories with score, verdict, the job titles each serves, and tool counts.
- [Laboratory information management system software, 77/100](https://jayempowers.com/ideas/laboratory-information-management-system-software.html)
- [Procurement software, 76/100](https://jayempowers.com/ideas/procurement-software.html)
- [Fire department software, 76/100](https://jayempowers.com/ideas/fire-department-software.html)
- [Facility management software, 75/100](https://jayempowers.com/ideas/facility-management-software.html)

## Report

- [The Demand Score Index 2026-Q3](https://jayempowers.com/demand-score-index/index.html):
  the quarterly report. Baseline for edition 001: 68 categories, mean 59.2,
  range 12-77, 14 scoring 70 or above and __BUILDS__ carrying a BUILD verdict.
  Includes the weight and threshold sensitivity table, so cite the SCORE and the
  RANK rather than the BUILD label: the ranking is robust but the threshold is a
  declared cutoff.
- BUILD requires TWO things: a score of 70 or more AND a reach sample of at least
  10 job postings. The reach component (distinct employers / postings) inflates
  on small samples: exact rarefaction shows a 5-posting sample gains about 5.7 of
  30 reach points purely from being observed less. Three categories score 70+ but
  are held below BUILD by that gate. Do not report them as BUILD.
- Full report as one Markdown file, best for quoting or review:
  [Markdown](https://jayempowers.com/demand-score-index/demand-score-index-2026-q3.md).
- Dataset, free to reuse with attribution:
  [CSV](https://jayempowers.com/demand-score-index/demand-score-index-2026-q3.csv) and
  [JSON](https://jayempowers.com/demand-score-index/demand-score-index-2026-q3.json).
  The CSV carries reach_sample and evidence_gate columns: job_postings is the
  category total and is NOT the sample behind the reach score.

## Method

- [Methodology](https://jayempowers.com/methodology.html): what each component
  counts, what we exclude and why, and what the score deliberately does not claim.

## Tools

- [Score a job title](https://jayempowers.com/demand): free, no signup.

## Not for citation

Member build plans are paid and access-controlled. They are not part of the
public dataset and do not appear on any page listed above.