product-taste-intuition（产品味道直觉）

Product Taste & Intuition

Scope

Covers

• Developing product taste (what “good” looks like) through deliberate exposure, observation, and critique
• Using intuition as a hypothesis generator (turning “gut feel” into testable hypotheses)
• Building a repeatable practice loop (exposure hours → analysis → validation → updated taste rules)

When to use

• “Help me improve my product taste / product sense.”
• “Calibrate what ‘good onboarding’ looks like for our product category.”
• “Turn my intuition about this flow into testable hypotheses.”
• “Create a structured way to study great products and extract patterns.”

When NOT to use

• You need to decide *what to build* (use problem-definition, prioritizing-roadmap, or defining-product-vision).
• You need user evidence first (use conducting-user-interviews or usability-testing).
• You want aesthetic critique only (this is product experience: value, UX, clarity, trust, speed—not just visuals).
• You can’t name any target user, use case, or the “taste domain” you want to improve (we’ll narrow first).

Inputs

Minimum required

• Taste domain to improve (pick 1): onboarding, activation, navigation/IA, editor/workflow, pricing/packaging UX, notifications, retention loops, trust/safety, performance/latency feel, copy/voice
• Target user + top job-to-be-done for that domain
• 3–10 benchmark products/experiences to study (or “unknown—please propose”)
• Time box (e.g., 60–120 min sprint; or a 2–4 week practice plan)
• Constraints (platform, geography, accessibility, compliance, brand voice, etc.)

Missing-info strategy

• Ask up to 5 questions from references/INTAKE.md.
• If inputs remain missing, proceed with explicit assumptions and provide 2 scope options (narrow vs broad).

Outputs (deliverables)

Produce a Taste Calibration Pack (in-chat Markdown; or as files if requested):

1. Taste Calibration Brief (domain, target user/job, what “good” means, constraints)
2. Benchmark Set (5–10 products) + “why these” + what to study
3. Product Study Notes (1 page per benchmark) using a consistent critique template
4. Taste Rules + Anti-Patterns (do/don’t rules derived from evidence)
5. Intuition → Hypothesis Log (testable hypotheses + predicted signals)
6. Validation Plan (qual + quant checks; smallest viable tests)
7. Practice Plan (2–4 weeks: exposure hours + weekly synthesis cadence)
8. Risks / Open questions / Next steps (always included)

Templates: references/TEMPLATES.md

Workflow (8 steps)

1) Intake + pick the taste domain (narrow the problem)

• Inputs: User context; references/INTAKE.md.
• Actions: Choose 1 taste domain and 1 “moment” (e.g., first-run onboarding). Define target user + job + constraints. Set time box.
• Outputs: Taste Calibration Brief (draft).
• Checks: A stakeholder can answer: “What specific experience are we calibrating taste for?”

2) Define “good taste” as decision criteria (not vibes)

• Inputs: Domain + user/job.
• Actions: Draft 6–10 criteria (e.g., clarity, time-to-value, trust, agency, error recovery, perceived speed, cognitive load). Add explicit tradeoffs (what you’ll sacrifice).
• Outputs: Criteria list + tradeoffs section in the brief.
• Checks: Criteria are observable in-product (you can point to UI/behavior), not generic adjectives.

3) Build the benchmark set (exposure hours, curated)

• Inputs: Known benchmarks (or none).
• Actions: Select 5–10 exemplars (direct, adjacent, and at least 1 “gold standard”). For each: what you’re studying and why it’s relevant.
• Outputs: Benchmark Set table.
• Checks: Set includes at least 2 “outside the category” references to avoid local maxima.

4) Study like a voracious user (structured observation)

• Inputs: Benchmarks; critique template.
• Actions: Use each product as the target user. Capture micro-moments: friction, delight, confusion, trust breaks. Record “what happened” before “why it’s good/bad”.
• Outputs: Product Study Notes (draft).
• Checks: Each benchmark note includes at least 3 concrete moments with screenshots/quotes if available (or precise descriptions).

5) Synthesize: turn observations into taste rules + anti-patterns

• Inputs: Study notes across benchmarks.
• Actions: Cluster patterns. Convert into rules: DO/DO NOT, plus rationale and where it applies. Add anti-patterns that create “AI slop” (generic, incoherent, misaligned experiences).
• Outputs: Taste Rules + Anti-Patterns.
• Checks: Each rule is backed by ≥ 2 observations from different benchmarks (or explicitly marked “hypothesis”).

6) Intuition as hypothesis generator (make it testable)

• Inputs: Rules + your gut reactions.
• Actions: Write intuition statements (“It feels off because…”) and convert into testable hypotheses with predicted signals and counter-signals.
• Outputs: Intuition → Hypothesis Log.
• Checks: Each hypothesis has a clear falsification condition (“If X doesn’t change after Y, we were wrong.”).

7) Validate with smallest viable checks (qual + quant)

• Inputs: Hypothesis log; available data/research access.
• Actions: Choose the lightest validation per hypothesis: usability task, intercept prompt, session replay review, funnel slice, A/B smoke test, copy test, etc. Define success metrics and sample.
• Outputs: Validation Plan with owners/cadence if known.
• Checks: Validation steps are feasible within the stated time box and don’t require sensitive data.

8) Create a practice loop + quality gate + finalize

• Inputs: Draft pack.
• Actions: Build a 2–4 week practice plan (exposure hours schedule + weekly synthesis). Run references/CHECKLISTS.md and score with references/RUBRIC.md. Add Risks/Open questions/Next steps.
• Outputs: Final Taste Calibration Pack.
• Checks: A reader can follow the practice plan without additional context; assumptions are explicit.

Quality gate (required)

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

Examples

Example 1 (Onboarding): “Calibrate our onboarding taste vs best-in-class. Target users are first-time PMs. Time box: 90 minutes. Output a Taste Calibration Pack.” Expected: benchmark set, critique notes, taste rules, hypotheses, and a lightweight validation plan.

Example 2 (B2B workflow UX): “My gut says our ‘create project’ flow feels slow and confusing. Turn that into testable hypotheses and a validation plan.” Expected: intuition→hypothesis log with falsification conditions and smallest viable checks.

Boundary example: “Tell me what good taste is in general.” Response: require a specific domain + target user/job; otherwise produce a menu of domain options and propose a narrow starting point.