Library
Talk to the Elephant: Design Learning for Behavior Change · 7 of 13
Talk to the Elephant: Design Learning for Behavior Change
learning-design CRITICAL

Implementation Playbook: From Diagnosis to Deployment

Talk to the Elephant: Design Learning for Behavior Change Julie Dirksen
implementation diagnostic-interview com-b-process practitioner-guide case-study

Key Principle

The behavioral design process follows a fixed diagnostic sequence. Skipping or reordering steps produces interventions aimed at the wrong target. The sequence:

Phase 1: Diagnose

  1. Specify the target behavior (Picture/Video Test)
  2. Run COM-B analysis (research, not assumption)
  3. Apply APEASE to prioritize behaviors if list is long
  4. Apply the Big Red Line Test: can the environment be fixed instead?
  5. Determine if training is actually the correct intervention

Phase 2: Map 6. Map COM-B gaps to Intervention Types (BCW) 7. Select specific BCTs within each intervention type 8. Identify where on the Change Ladder your audience is 9. Design for the full Learning Journey (not just the training event)

Phase 3: Design 10. Build value communication for the Elephant (Significant, Immediate, Tangible, Likely) 11. Design practice for automaticity, not just accuracy 12. Build implementation intentions into the design 13. Design environmental/social supports alongside the training 14. Apply ethical checklists (research, strategy, design phases)

Phase 4: Test and Iterate 15. Test with real learners (not SMEs) 16. Measure behavioral outcomes, not just satisfaction or knowledge scores 17. Deploy BCT 4.4 (Behavioral Experiments) post-implementation 18. Design for Maintaining stage — what happens after the training event?

How to Run a COM-B Diagnostic Interview

COM-B analysis requires research with the target audience. Do not diagnose from assumptions. Sources:

  • Interviews with representative performers (positive deviants, average performers, non-performers)
  • Observation of the behavior in its natural context
  • Stakeholder interviews (managers, adjacent roles who observe the behavior)
  • Data review (performance records, audit trails, incident reports)

Interview coding (from the Truth Initiative smoking cessation case, Chapter 15): After conducting behavioral interviews, code each barrier/enabler statement to a COM-B sub-factor:

  • "I don't know how to..." → Psychological Capability
  • "I don't have time / the tools aren't available" → Physical Opportunity
  • "My colleagues don't..." → Social Opportunity
  • "I don't feel like it matters" → Reflective Motivation
  • "It's just not my habit" → Automatic Motivation

The pattern of codes across interview subjects reveals which sub-factors are most prevalent and most design-amenable.

The Chapter 14 Case Study: Complete Diagnostic Chain

Context: Jeong (L&D), Anisha (design lead), and Rita (accessibility advocate) are building an intervention to improve alt text compliance in a design team.

COM-B Analysis:

  • Psychological Capability gap: Most designers know accessibility is important but don't know the specific alt text standard (what good looks like).
  • Physical Opportunity gap: The publishing workflow has no alt text field; compliance requires adding a manual step outside the existing process.
  • Social Opportunity gap: No peer norm around accessibility; it is not discussed in design critiques or seen as a quality criterion.
  • Automatic Motivation gap: No habit — the behavior has never been performed consistently enough to become automatic.
  • Reflective Motivation: Partially present (designers care about inclusivity in principle) but not linked to alt text specifically.

Intervention types selected:

  • Education (Psychological Capability gap): BCT 4.1 Instruction on standard
  • Environmental Restructuring (Physical Opportunity gap): BCT 12.1 Add alt text field to publishing form
  • Modeling (Social Opportunity gap): BCT 6.1 Demonstrate, BCT 13.1 Identify role models on team
  • Training (Automatic Motivation gap): BCT 8.3 Habit formation through repeated practice in workflow
  • Persuasion (Reflective Motivation gap): BCT 5.1 Salience — show the user experience of screen reader users

Learning Journey design:

  • Prelearning: Self-assessment of current alt text quality in their own recent work
  • Learning: Standard and examples
  • Practice: Critique session on existing materials + implementation in current projects
  • Job Aid: Alt text decision tree in the publishing form
  • Refreshing: Monthly accessibility review incorporated into design critique rotation
  • Developing Further: Accessibility champion role created (BCT 13.1)

Common Practitioner Errors (Ten Most Frequent)

  1. Prescribing training before diagnosing root cause. Training is one of nine intervention types. It addresses only Capability.
  2. Diagnosing from assumptions, not research. "I know what the problem is" is not COM-B analysis.
  3. Specifying outcomes rather than behaviors. "Improve patient satisfaction" is not a behavior. "Greet patient by name within 30 seconds of entering room" is.
  4. Designing for the average learner. Change Ladder position varies widely. Design for a specific audience segment at a specific stage.
  5. Treating motivation as monolithic. Reflective and Automatic Motivation are different systems requiring different BCTs.
  6. Designing for the training event and nothing else. The Learning Journey has seven categories; the event covers at most two.
  7. Using expected rewards for intrinsically motivated behaviors. The Overjustification Effect degrades intrinsic motivation.
  8. Running SME review instead of learner testing. SMEs cannot simulate learner experience.
  9. Using behavioral contracts (BCT 1.8/1.9) in mandatory contexts. Imposed accountability triggers reactance.
  10. Measuring satisfaction or knowledge scores as behavioral outcomes. Kirkpatrick Level 1-2 data says nothing about Level 3-4 behavior change.

Practitioner Field Examples (Chapter 15)

Truth Initiative — BecomeAnEx smoking cessation app:

  • COM-B interviews coded to sub-factors before any design decision
  • Stage-matched content tracks — Contemplators received different content than Maintainers
  • Usability testing before efficacy testing: "If they can't use it, it doesn't matter what it contains"
  • Iterative prototyping with real users, not focus groups of stakeholders

Female entrepreneurship curriculum (low-literacy international contexts):

  • Three-stage materials testing: existing materials first, then modified, then new
  • 50/30/20 image rule as design constraint
  • Buddy system as BCT 3.2 — structured peer support with shared task focus
  • Facilitator training as a testing environment, not just delivery preparation

Valor Nigeria HIV testing:

  • Anonymous WhatsApp co-research to reach the audience without disclosure risk
  • Affirmation-first onboarding before any behavior-change ask
  • Valor (courage in service of community) as the reframe — existing value, new behavioral expression
  • Result: male testing rates shifted from ~1/3 to ~1/2 of new testers

Rules of Thumb

  • The diagnostic sequence is: behavior → COM-B → intervention types → BCTs. Never reverse this.
  • Every BCT selection must be traceable to a specific COM-B gap. If you can't make that trace, drop the BCT.
  • Test with three learners before any major production investment. Three learners will reveal 70-80% of usability and comprehension problems.
  • Design explicit Maintaining-stage support. Most initiatives fund the training event; almost none fund the Maintaining phase. This is where durable change fails.
  • Measure behavior, not just learning. If you can't measure the behavior, measure the closest behavioral proxy — and be explicit about what that proxy does and doesn't tell you.

Related References