How we measure
Every number on this site comes from a recorded source. Here is where each one comes from and when it was measured.
| Number on site | Source | Measured |
|---|---|---|
| 82% of work items needed no rework | 51 of 62 measured stories in one complete delivery cycle of our engine | Sep 2026 |
| 61 of 62 sized correctly | Same cycle; the one miss was a human override | Sep 2026 |
| 1,200 tests, 0 failing | Marketplace build, test suite on the main branch | Aug 2026 |
| 7 of 7 recurring defects caught by independent review | Learning report of the same cycle: all 7 occurrences of the two most frequent defect classes | Sep 2026 |
| ~3,000 tests (case study 1) | Gateway ~1,150 · harness ~1,070 · evidence tooling ~760 · infrastructure 75 | 7 Oct 2026 |
| €1,479 vs €12.99 (case study 2) | Defect reproduced and fixed during the code review | Aug 2026 |
Figures are measurements on the dates shown; they are refreshed quarterly.
Find out what needs fixing before you invest further.
You get a report that stays yours, whether or not we work together afterwards.