Case selection

The corpus pairs one hundred success cases (S-series) with one hundred failure cases (F-series). Each case is an episode, not a company biography: one organisation, one bounded period, and one strategic question the episode turned on. Cases span industries, geographies, and eras — from regulated banks and industrial manufacturers to public-sector agencies — and each is assigned an archetype describing the kind of episode it is.

Sourcing standards

Every case is reconstructed from cited sources ranked in three tiers: primary (filings, court records, first-person accounts, contemporaneous corporate documents), secondary with stated justification (scholarly case studies, investigative journalism), and tertiary sources, which are flagged as such and never load-bearing. Claims in the case narrative are attributed to specific sources, and source-level verification is an ongoing, audited process — errors found in it flow into the Corrections Log.

The scoring grid

Each case is scored on the fifteen-cell Success Architecture Grid: the three OTA phases (Observe, Think, Act) crossed with the five modalities (Direction, Structure, Processes, Capability, Culture). Scoring distinguishes root-cause phases from transmission steps — a phase can perform well and still carry a failure downstream — and modalities carry explicit weights, with a zero-modality rationale recorded whenever an evidenced modality is judged to carry no independent causal weight.

Quality control

Cases are scored under a blind review contract: raters score from the case file’s episode summary, sources, and OTA narrative against the methodology document, without access to other raters’ work or prior internal analyses. A separate QC pass audits scored cases for methodology compliance, and a second, independent audit verifies that cited sources actually support the claims made.

Versioning

The methodology is versioned; the corpus published here is scored under version 4. Corrections that change a case record or a score are logged publicly, and material methodology changes increment the version rather than silently rewriting history.

[TO BE PUBLISHED] The full methodology write-up — scoring rubrics, evidence rules, weighting scheme, and worked scoring examples — will be published on this page, derived from the canonical versioned scoring document.