Organizational development, with the evidence stated
A comprehensive guide for knowledge workers — health care professionals, software engineers, and senior executives. Every framework says what it is, when to use it, when not to, and how strong the evidence behind it actually is.
What this is
One guide you can read end to end, and one reference file per model behind it.
The guide
Eleven parts, from foundations and the OD process through diagnosis, change, culture, teams, interventions, measurement, and anti-patterns. Self-contained: you can act without following a single link.
The topics
88 reference files, one per model, framework, method, and system — each with its origin, its structure, worked examples for all three audiences, and its limitations.
The questionnaires
Every diagnostic model is paired with an operational instrument: numbered steps, rating scales, and a closing analysis section that says how to read the scores.
Claude Code skills
Two skills ship in skills/ — one applies the guide to a real situation, the other maintains this repository per its own conventions.
The evidence, counted
Most OD writing presents every framework as equally sound. This collection does not.
Each model carries an evidence label with a specific, named objection rather than a generic hedge. Across 72 models the distribution is unflattering to the field — and that is the accurate picture. A weak model can still be a useful vocabulary; the label says which job it can and cannot do. Questionnaires are not counted here: each one inherits the rating of the model it operationalizes.
| Strength | Topics | What it means |
|---|---|---|
| Strong | 5 | Meta-analytic or large replicated experimental support. |
| Good | 8 | Substantial empirical support, with some methodological limits. |
| Moderate | 12 | Real support for the underlying principle; the model itself untested. |
| Mixed | 6 | Genuinely contested, or components differ sharply in support. |
| Contested | 1 | Influential and seriously criticized on method. |
| Weak | 30 | Practitioner-derived, with little or no independent validation. |
| Very weak | 3 | Contradicted, or transferred from an unrelated domain. |
| Not an empirical model | 3 | A taxonomy or checklist, to be judged on usefulness. |
| Not a model | 3 | A method or guidance, to be judged on how well it is run. |
| Normative | 1 | Value commitments, to be argued rather than measured. |
Where to start
Seven families of material. Diagnose before you change; measure before you claim.
Foundations, process, and ethics
what OD is, how an engagement runs, and the obligations that come with it
10 topics
- Conway's law Good
- Action research Moderate
- Sociotechnical systems Moderate
Diagnostic models
frameworks for working out what is actually wrong, each with a questionnaire
28 topics
- Organizational network analysis Good
- Nadler-Tushman congruence model Moderate
- Galbraith star model Moderate
Change models
how change is planned, sequenced, and lived through
13 topics
Culture models
what culture is, how to read it, and what actually moves it
6 topics
Team and individual models
the behaviour of groups and the motivation of the people in them
20 topics
- Psychological safety Strong
- Self-determination theory Strong
- Goal setting theory Strong
Facilitation and large-group methods
ways to convene many people and come out with something usable
6 topics
Measurement
what to count, what counting costs, and what it cannot tell you
5 topics
- DORA metrics Good
- SPACE framework Moderate
- High reliability organizations Moderate
The guide, part by part
Read it in order, or jump to the part you need.
Colophon
The prose lives in the source repository and is vendored into this site, which is built with SvelteKit and the Lily Design System™ and served as static files by GitHub Pages. Corrections and additions are welcome as pull requests against the source.