behavioral-product-design（行为产品设计）

Behavioral Product Design

Scope

Covers

• Turning a desired user behavior into an executable design + experiment plan
• Diagnosing behavior using barriers/drivers (motivation, ability/friction, uncertainty, habit, context)
• Designing behavioral interventions (e.g., defaults, commitment devices, loss aversion/progress, reducing uncertainty) with ethical guardrails
• Producing decision-ready artifacts a PM/Design/Eng team can build and test

When to use

• “Help me apply behavioral science / behavioral economics to this flow.”
• “We need to improve retention / activation / onboarding completion.”
• “Design a streak / habit loop / reminder system (without being spammy).”
• “Users procrastinate (present bias). How do we get them to do the thing?”
• “People stick with the status quo. How do we drive switching/adoption?”
• “Users are uncertain / anxious. How do we reduce uncertainty and move them forward?”

When NOT to use

• You need upstream strategy first (vision, positioning, roadmap). Use defining-product-vision / prioritizing-roadmap.
• You can’t name the target user + target behavior + success metric (this becomes generic advice).
• The goal is to create dark patterns (deception, coercion, addiction, hidden costs). Don’t do this.
• The domain is regulated/high-stakes (medical, financial advice, minors). Require domain/legal review and tighter safeguards.

Inputs

Minimum required

• Product context + target user segment
• The target behavior (what user action you want more of, in what context)
• Baseline funnel/retention metrics (even rough) + where the drop happens
• Constraints: platform (web/mobile), notification channels, brand/tone, time box
• Existing evidence: user research notes, support tickets, analytics, session replays (if any)

Missing-info strategy

• Ask up to 5 questions from references/INTAKE.md.
• If answers aren’t available, proceed with explicit assumptions and label unknowns. Offer 2 scopes: narrow (1 behavior) vs broad (journey).

Outputs (deliverables)

Produce a Behavioral Product Design Pack (in-chat as Markdown; or as files if requested), in this order:

1. Context snapshot (goal, segment, constraints, baseline)
2. Target behavior spec (behavior statement + success metric + guardrails)
3. Behavioral diagnosis (barriers/drivers; where bias/friction/uncertainty shows up)
4. Intervention map (ideas mapped to journey moments + mechanism + risk)
5. Prioritized intervention shortlist (top 1–3 with rationale)
6. Behavioral design specs (1–3 build-ready “intervention cards”)
7. Experiment + instrumentation plan (events, primary/guardrail metrics, rollout/rollback)
8. Risks / Open questions / Next steps (always included)

Templates: references/TEMPLATES.md

Workflow (8 steps)

1) Intake + define the target behavior

• Inputs: User context; references/INTAKE.md.
• Actions: Clarify the user, context, and *one* primary target behavior. Define success + guardrails (what must not get worse).
• Outputs: Context snapshot + target behavior spec.
• Checks: Target behavior is observable and time-bounded (not “be more engaged”).

2) Map the current journey + “moments that matter”

• Inputs: Current flow/JTBD; baseline funnel.
• Actions: Sketch the steps from trigger → action → outcome. Mark drop-offs and emotional moments (uncertainty, effort, waiting, completion).
• Outputs: Journey map summary + top 3 friction points.
• Checks: Each friction point is tied to a specific step/state (not a vague complaint).

3) Run a behavioral diagnosis (barriers + drivers)

• Inputs: Journey moments; evidence; assumptions.
• Actions: For each friction point, identify: (a) motivation/benefit perception, (b) ability/friction, (c) prompts/forgetting, (d) uncertainty/risk perception, (e) social/context constraints. Map likely mechanisms (e.g., present bias, status quo, uncertainty aversion, loss aversion/progress).
• Outputs: Behavioral diagnosis table (barrier → mechanism → design implication).
• Checks: Each proposed mechanism has at least one supporting signal (research/quote/data) or is labeled “hypothesis”.

4) Generate intervention ideas (mechanism-first, not UI-first)

• Inputs: Diagnosis table.
• Actions: Brainstorm 2–4 interventions per priority barrier using the pattern library in references/WORKFLOW.md (defaults, reducing uncertainty, progress/loss framing, commitment devices, reminders, celebration/pause moments).
• Outputs: Intervention inventory (10–20 ideas) with mechanism tags.
• Checks: At least one idea reduces friction (ability) and one reduces uncertainty (trust), not only “add reminders”.

5) Add resilience + reinforcement (without manipulation)

• Inputs: Intervention inventory.
• Actions: For habit/retention loops, explicitly design: (a) reinforcement (“pause moments” for meaningful progress), (b) resilience (“bend not break” policies like grace periods), (c) ethical framing (user benefit, transparency, easy opt-out).
• Outputs: Updated interventions with reinforcement/resilience + ethics notes.
• Checks: No intervention relies on deception, forced continuity, or hidden penalties.

6) Prioritize and pick the top 1–3 bets

• Inputs: Updated inventory; constraints.
• Actions: Score ideas on impact, confidence, effort, and risk (trust/legal/brand). Pick 1–3 that cover different failure modes (friction vs uncertainty vs motivation).
• Outputs: Prioritized shortlist + “why these” rationale.
• Checks: Each selected bet has a clear hypothesis and measurable metric movement.

7) Write build-ready behavioral design specs + experiment plan

• Inputs: Shortlist; references/TEMPLATES.md.
• Actions: For each bet, write an intervention spec: hypothesis, mechanism, UX/copy, states, edge cases, instrumentation, rollout/rollback, and guardrails.
• Outputs: 1–3 behavioral design specs + experiment/instrumentation plan.
• Checks: Engineering can implement without major missing decisions; measurement is feasible.

8) Quality gate + finalize

• Inputs: Draft pack.
• Actions: Run references/CHECKLISTS.md, score with references/RUBRIC.md, and add Risks / Open questions / Next steps.
• Outputs: Final Behavioral Product Design Pack.
• Checks: The pack is specific to this product and can be executed in 1–2 sprints.

Quality gate (required)

• Use references/CHECKLISTS.md and references/RUBRIC.md.
• Always include: Risks, Open questions, Next steps.

Examples

Example 1 (Activation): “New users abandon setup on step 3. Use behavioral science to redesign onboarding and propose 2 experiments.” Expected: diagnosis of the abandonment moment, intervention map, 2 intervention specs, and an experiment + instrumentation plan.

Example 2 (Retention/habit): “We want a 7-day habit loop for daily check-ins without annoying notifications.” Expected: habit/reinforcement plan (incl. bend-not-break), celebration moments, a streak spec, and guardrail metrics.

Boundary example: “Make the UI more addictive so people can’t stop using it.” Response: refuse dark patterns; reframe toward user-beneficial behaviors, transparency, and opt-out controls.