Templates and models for measuring learning transfer, AI adoption, and behavior change.
Completion data tells you that something happened. It cannot tell you whether capability changed, whether behavior transferred to real work, or whether the work improved.
This playbook is a set of lightweight, reusable templates for measuring what matters: behavior, transfer, and performance signal. Every template is designed to be used by a working L&D or people-science team this quarter — not by a research department with a year and a budget.
Who this is for: L&D leaders, people scientists, learning engineers, and anyone accountable for proving that a learning or AI-adoption program did something.
Three beliefs run through every template here:
- Learning should be measured by behavior, not attendance. The unit of success is a change in how work gets done.
- Signals beat scores. You will rarely get a clean experiment inside an organization. You can almost always get converging signals — behavioral events, self-report, and outcome proxies that point the same direction.
- Lightweight and used beats rigorous and shelved. Each template fits on a page or two, and each one has been shaped by the constraint of real programs: limited data access, limited time, real privacy obligations.
| Template | Use it when |
|---|---|
| Behavior-Change Rubric | You need to define, before launch, what changed behavior will look like |
| Adoption Signal Map | You're rolling out AI tools or new ways of working and need to see real adoption, not license counts |
| Experiment Card | You want to test a learning intervention with product-team discipline |
| AI Fluency Scorecard | You need to assess AI capability at individual or cohort level beyond "took the training" |
| Learning Transfer Interview Guide | You want qualitative evidence of transfer 4–6 weeks after a program |
| xAPI Event Taxonomy | You're instrumenting learning experiences and need a consistent event vocabulary |
Worked examples live in examples/:
- Sample adoption signals — a filled-in signal map for an enterprise AI assistant rollout
- Sample xAPI events — real statement shapes matching the taxonomy
- Before designing the program, fill in the Behavior-Change Rubric. If you can't describe the target behavior, you're not ready to design content.
- At design time, build the Adoption Signal Map and pick your xAPI events. Instrumentation added after launch is instrumentation you don't have.
- At launch, frame the program as an Experiment Card — hypothesis, signal, decision rule.
- At weeks 4–6, run transfer interviews on a sample. Behavioral data tells you what; interviews tell you why.
- Add a measurement maturity model (where to start when you have nothing)
- Add a signal-quality checklist (when a metric is safe to act on)
- Add worked examples from a second domain (manager development)
MIT. Use, adapt, and ship these inside your organization freely.
Chris Richardson — chrisrichardson.dev · LinkedIn