Evidence
Measured results from real teams · Last updated 2026-07-11 · Ousios LLC
This page holds Composa's published outcome evidence: measured before/after results from design-partner teams, including what didn't work. It exists so that when we have proof, it lives at a permanent address — and so that until we do, nobody has to wonder.
0
Zero case studies published. The design-partner program is in progress — five teams, measured on metrics they choose, with results published whatever they show. Until then, our limitations page says it plainly: plausibility is not efficacy, and we won't imply evidence we don't have.
What every published case study will contain
| The team | Size, role mix, and industry — named, industry-only, or anonymized at the team's choice; the team approves every word before publication. |
| The question | The 2–3 success metrics the team chose before starting, with the pass/fail thresholds they wrote down on day 0. |
| The method | How the baseline and after-measurements were taken (self-timed windows per the pilot protocol), and over what period. |
| The numbers | Baseline vs after, per metric — actual values, not percentages without denominators. |
| What didn't work | The workflows that got swapped or abandoned, and why. If this section is thin, be suspicious of the rest. |
| The team's verdict | In their words, unedited beyond length. |
Why publish failures? Because one honest case study with a visible failure section is worth more to a skeptical buyer than fifty five-star quotes — and because our own failure-modes page committed us to this standard before any results existed.