Run a Behavioral Strategy workshop #
Lead a session in which participants choose a target behavior, compare alternatives, and decide what to test. Use the M-PESA case for a first session. Instagram and GOV.UK Verify are alternatives for a group whose work is closer to consumer products or public services.
This is a proposed 90-minute teaching design. The timings below are planning estimates, not observed results from independent delivery. Prior knowledge of the framework is optional. A facilitator should read the case, its worked analysis, and the teaching notes before the session.
Prepare the session #
Download the complete pack, or use the individual files. Give participants the student casebook and a blank decision workbook. Keep the facilitator guide separate until the discussion reaches the comparison stage. The slides are optional and include speaker notes.
Plan for individuals or small groups who can each produce one recommendation. Give each group paper or an editable workbook. Provide the case in advance if readers need more time or another format. Participants can work through the same material without a projector or an account.
Write this outcome where everyone can see it: By the end, each group will recommend a bounded next step, explain its evidence, and name a finding that would change the decision.
The 90-minute agenda #
| Minutes | Activity | Facilitator action | Participant output |
|---|---|---|---|
| 0-8 | Frame the decision | Explain outcome, behavior, and feature. State that the case does not document historical use of the framework. | One sentence describing the decision |
| 8-20 | Read the case | Open only the student packet. Ask readers to mark a fact, an inference, and an unknown. | Three labelled notes |
| 20-32 | Generate alternatives | Use workbook steps 1-2. Require an option that reduces the user’s work, where plausible. | Candidate comparison |
| 32-45 | Map the delivery chain | Use steps 3-4. Ask whose action and support each option needs. | Actor chain with evidence gaps |
| 45-60 | Design the test | Use steps 5-6. Ask for the opportunity denominator, repeat window, acceptable support, and stopping rules. | A test plan that someone could run |
| 60-70 | Commit to a recommendation | Use steps 7-8. Ask each group to state its decision before the worked analysis opens. | Recommendation and reversal condition |
| 70-82 | Compare the reasoning | Read the worked analysis. For M-PESA, compare both responses. Use the discussion rubric below. | One retained choice and one revision |
| 82-90 | Transfer the method | Apply the last prompt to a different, nonconfidential setting. | A new candidate and a measurement question |
The table totals 90 minutes. Longer reading or discussion requires a longer session or advance reading; do not compress the test plan to preserve every presentation slide.
Prompts at the difficult points #
When a group names a feature, ask: “Who does what, in which situation, and how does that action contribute to the outcome?” When it names an outcome, ask what observable action would have to happen first.
When a group chooses its favorite option immediately, ask for the strongest alternative and an option that removes a task from the user. Removing work can transfer work to an agent, operator, employer, or government team; put that actor on the chain.
When a group treats interest as evidence, ask what people actually did, how they were selected, and what help they received. When it treats repeated activity as automatic proof, ask whether the person had another opportunity and whether the action served the intended outcome.
When a group proposes a test, ask another group to explain how it would count success and failure. If the second group cannot tell, revise the record before discussing sample size.
Discuss the recommendation #
Use the six-part rubric. A defensible answer has a clear action and population, serious alternatives, an actor chain, traceable evidence, an executable test, and a proportionate next decision. Agreement with the historical outcome earns no special credit.
Invite each group to state one disagreement with the worked response and the evidence needed to resolve it. A well-supported decision to stop or collect missing evidence can be stronger than a confident proposal to expand.
Transfer exercise #
Choose a routine process that can be described without private company or personal information: booking a meeting, returning equipment, or finding an internal document. State the outcome. Generate two candidate behaviors and one option that removes user work. Identify a dependent actor and one observation that could disqualify the preferred option. This is a planning exercise; it does not authorize collecting personal data or running a live study.
After the session #
Let participants keep and revise their work. If you want to assess the teaching material, use the optional host observation record. Record what needed explanation and what the group could do unaided. Do not infer a learning effect or method validation from satisfaction ratings.
Original teaching content: Jason Hreha, BehavioralStrategy.com, teaching pack v1.0 (2026). CC BY 4.0.