Files
gh-lyndonkl-claude/skills/discovery-interviews-surveys/resources/evaluators/rubric_discovery_interviews_surveys.json
2025-11-30 08:38:26 +08:00

266 lines
16 KiB
JSON

{
"criteria": [
{
"name": "Research Objective Clarity",
"description": "Is the research objective clearly defined with specific learning goals and hypotheses?",
"scoring": {
"1": "Vague objective ('learn about users'). No specific learning goals. Hypotheses missing. No connection to decision.",
"3": "Objective stated but could be more specific. Some learning goals identified. Hypotheses present but may lack precision. Decision somewhat clear.",
"5": "Crystal clear objective with specific learning goals. Testable hypotheses documented. Clear connection to decision to be informed. Success criteria defined."
}
},
{
"name": "Participant Targeting & Recruitment",
"description": "Are target participants well-defined with appropriate screening and recruitment strategy?",
"scoring": {
"1": "Participant criteria vague or missing. No screening questions. Recruitment strategy not specified. Sample size not justified.",
"3": "Participant criteria identified but may lack specificity. Basic screening questions. Recruitment strategy mentioned. Sample size roughly appropriate.",
"5": "Precise participant criteria (demographics, behaviors, firmographics). Comprehensive screening questions. Clear recruitment strategy with channels. Sample size justified by method (5-15 for qual, 100+ for quant, 30+ per segment)."
}
},
{
"name": "Question Quality & Bias Avoidance",
"description": "Are questions well-designed, open-ended, behavior-focused, and free of leading bias?",
"scoring": {
"1": "Many leading or hypothetical questions ('Would you...?', 'Don't you think...?'). Closed yes/no questions dominate. No behavior focus. Confirmation bias obvious.",
"3": "Mix of open and closed questions. Some behavior focus but also hypotheticals. Minor leading language. Some bias-avoidance techniques attempted.",
"5": "Exemplary questions: open-ended, behavior-focused ('Tell me about the last time...'), use 'show me' requests, avoid hypotheticals, no leading language. Systematic bias-avoidance techniques applied throughout."
}
},
{
"name": "Interview Guide / Survey Structure",
"description": "Is the research instrument well-structured with logical flow and appropriate depth?",
"scoring": {
"1": "Poor structure. No logical flow. Too shallow (only surface questions) or too narrow (missing key areas). Inappropriate question types.",
"3": "Decent structure with some flow. Covers main topics but may miss areas. Question types mostly appropriate. Could use refinement.",
"5": "Excellent structure: warm-up, core questions, concept test (if applicable), wrap-up. Logical flow from general to specific. Appropriate depth and breadth. Question types match objectives (open-ended for discovery, scaled for validation). Includes follow-up probes."
}
},
{
"name": "JTBD / Problem Discovery Focus",
"description": "For discovery research, does it focus on jobs-to-be-done, problems, and context (not just features)?",
"scoring": {
"1": "Feature-focused ('Do you want feature X?'). No JTBD exploration. Missing context about problems, workflows, or triggers. Hypothetical focus.",
"3": "Some problem exploration. Brief JTBD elements. Context partially explored. Mix of problem and feature questions.",
"5": "Deep JTBD focus: hiring/firing triggers, desired outcomes, current workarounds, pain points, context. Timeline reconstruction for switchers. Problems before solutions. 'Show me' requests for workflows."
}
},
{
"name": "Sample Size & Statistical Rigor",
"description": "Is sample size appropriate for method and are statistical considerations addressed for surveys?",
"scoring": {
"1": "Sample size inappropriate (e.g., 3 interviews claiming statistical significance, or 20-person survey). No power analysis. No consideration of statistical validity for quantitative claims.",
"3": "Sample size roughly appropriate but not justified. Some statistical awareness for surveys (e.g., descriptive stats). May lack power analysis or significance testing.",
"5": "Sample size justified: 5-15 for qualitative themes, 100+ for survey stats, 30+ per segment for comparisons. Power analysis for surveys (margin of error, confidence level). Statistical tests specified (t-test, chi-square, etc.). Saturation check for interviews."
}
},
{
"name": "Analysis Plan & Rigor",
"description": "Is there a clear, systematic analysis plan with rigor techniques?",
"scoring": {
"1": "No analysis plan. Unclear how data will be processed. No mention of systematic approach, coding, or statistical tests. Risk of cherry-picking.",
"3": "Basic analysis plan. For interviews: thematic coding mentioned. For surveys: descriptive stats mentioned. Some structure but may lack rigor techniques.",
"5": "Comprehensive analysis plan. For interviews: systematic thematic coding, affinity mapping, frequency counting, saturation check, negative case analysis. For surveys: descriptive stats, inferential tests, segmentation, open-end coding. Pre-specified to avoid p-hacking."
}
},
{
"name": "Facilitation & Execution Guidance",
"description": "For interviews, is there guidance on facilitation techniques (active listening, probes, silence)?",
"scoring": {
"1": "No facilitation guidance. Script-only approach with no flexibility. Missing techniques like probing, silence, active listening.",
"3": "Some facilitation guidance. Probes included for some questions. Brief mention of techniques. Could be more comprehensive.",
"5": "Detailed facilitation guidance: active listening, follow-up probes ('Tell me more', 'Why did that matter?'), embrace silence (3-5 sec pause), mirroring, 'show me' requests, non-verbal cue awareness. Recording and note-taking protocol."
}
},
{
"name": "Ethics & Consent",
"description": "Are ethical considerations addressed (informed consent, privacy, compensation)?",
"scoring": {
"1": "No mention of consent, privacy, or compensation. Ethical considerations ignored.",
"3": "Brief mention of consent or compensation. Some privacy awareness. May lack detail on implementation.",
"5": "Comprehensive ethics: informed consent script, explicit recording permission, anonymization in reports, secure data storage, fair compensation specified, opt-out option. Privacy-first approach."
}
},
{
"name": "Insights Documentation & Actionability",
"description": "Is there a clear template or plan for documenting insights with evidence and recommendations?",
"scoring": {
"1": "No insights documentation plan. Unclear how findings will be communicated. Missing connection to actionable recommendations.",
"3": "Basic insights template. Some structure for documenting findings. Recommendations mentioned but may lack specificity. Somewhat actionable.",
"5": "Comprehensive insights document template: executive summary, methodology, key findings with evidence (quotes/stats), surprises, recommendations (specific actions), confidence level, limitations. Actionable and decision-ready."
}
}
],
"minimum_score": 3.5,
"guidance_by_research_type": {
"Problem Discovery Interviews": {
"target_score": 4.0,
"focus_criteria": [
"Question Quality & Bias Avoidance",
"JTBD / Problem Discovery Focus",
"Facilitation & Execution Guidance"
],
"sample_size": "8-15 participants",
"key_requirements": [
"Open-ended, behavior-focused questions",
"Focus on past behavior, not hypotheticals",
"'Show me' requests for workflows",
"Problem before solution",
"Current workarounds explored",
"Systematic thematic coding planned"
],
"common_pitfalls": [
"Asking 'Would you use...' instead of 'Tell me about the last time you...'",
"Jumping to solutions before understanding problems",
"Not probing deeply enough (stopping at surface answers)",
"Selection bias (only interviewing enthusiasts)"
]
},
"Jobs-to-be-Done Research": {
"target_score": 4.2,
"focus_criteria": [
"JTBD / Problem Discovery Focus",
"Question Quality & Bias Avoidance",
"Interview Guide / Survey Structure"
],
"sample_size": "10-15 recent switchers",
"key_requirements": [
"Recruit recent switchers (last 3-6 months)",
"Timeline reconstruction (first thought → current state)",
"Forces of progress (push, pull, anxiety, habit)",
"Hiring/firing triggers identified",
"Desired outcomes vs current capabilities",
"Context and constraints explored"
],
"common_pitfalls": [
"Interviewing people who switched too long ago (memory fades)",
"Not reconstructing timeline (missing trigger events)",
"Ignoring anxieties and habits (forces resisting change)",
"Focusing only on product features, not job to be done"
]
},
"Concept Testing (Qualitative)": {
"target_score": 3.8,
"focus_criteria": [
"Question Quality & Bias Avoidance",
"Interview Guide / Survey Structure",
"Participant Targeting & Recruitment"
],
"sample_size": "10-15 target users",
"key_requirements": [
"Comprehension check ('In your own words, what is this?')",
"Target audience validation ('Who is this for?')",
"Use case exploration ('When would you use it?')",
"Value perception (pricing anchors, comparisons)",
"Concerns and objections surfaced",
"Avoid leading ('Don't you think this is great?')"
],
"common_pitfalls": [
"Testing with wrong audience (not actual target users)",
"Leading participants to 'correct' answer",
"Not exploring concerns (only positive feedback)",
"Mistaking 'sounds interesting' for 'will actually use'"
]
},
"Quantitative Surveys": {
"target_score": 4.1,
"focus_criteria": [
"Sample Size & Statistical Rigor",
"Question Quality & Bias Avoidance",
"Analysis Plan & Rigor"
],
"sample_size": "100+ overall, 30+ per segment",
"key_requirements": [
"Sample size justified (power analysis, margin of error)",
"Mix of scaled questions and open-ends",
"Avoid leading language in questions",
"Randomize option order",
"Pilot test with 5-10 people",
"Statistical tests pre-specified (t-test, chi-square, etc.)",
"Segmentation plan for subgroup analysis"
],
"common_pitfalls": [
"Too small sample for statistical claims (n < 30 per segment)",
"Leading questions ('How much do you love our product?')",
"No pilot test (discovering issues after launch)",
"Cherry-picking significant results (p-hacking)",
"Ignoring non-response bias"
]
},
"Continuous Discovery": {
"target_score": 3.7,
"focus_criteria": [
"Interview Guide / Survey Structure",
"Facilitation & Execution Guidance",
"Insights Documentation & Actionability"
],
"sample_size": "3-5 conversations per week",
"key_requirements": [
"Lightweight process (15-30 min conversations)",
"Rotating team members (product, eng, design)",
"Shared notes repository",
"Flexible guide based on current focus",
"Monthly synthesis of patterns",
"Relationship building with customers"
],
"common_pitfalls": [
"Making it too formal (blocks adoption)",
"Only product team participates (team stays disconnected)",
"No shared documentation (insights lost)",
"No periodic synthesis (patterns missed)",
"Stopping after a few weeks (not continuous)"
]
}
},
"common_failure_modes": [
{
"failure": "Hypothetical questions ('Would you...?')",
"symptom": "Questions like 'Would you pay $X?', 'Would you use feature Y?', 'If we built Z, would you switch?'. Participants describe future intent, not past behavior.",
"detection": "Look for 'would', 'if', 'imagine'. Check if questions focus on actual past behavior vs hypothetical scenarios.",
"fix": "Reframe to past behavior: 'Tell me about the last time you [needed this]', 'What have you tried?', 'Show me your current workflow'. Use 'have you' not 'would you'."
},
{
"failure": "Leading questions (telegraphing desired answer)",
"symptom": "Questions like 'Don't you think...?', 'Isn't it true that...?', 'How much do you love...?'. Bias obvious.",
"detection": "Check if question suggests 'right' answer. Would neutral observer detect researcher's opinion from question?",
"fix": "Use neutral phrasing: 'What's your experience with...?', 'Walk me through...'. Let participant form own opinion, don't guide."
},
{
"failure": "Insufficient sample size for claims",
"symptom": "Statistical claims from tiny samples ('80% of users want X' from 5 interviews). Surveys with n < 30 per segment claiming significance.",
"detection": "Check sample size vs type of claim. Interviews → themes only. Surveys → need n ≥ 30 per segment for stats.",
"fix": "For interviews: Report themes ('8/12 mentioned Y'), not percentages. For surveys: Ensure n ≥ 100 overall, 30+ per segment. Run power analysis."
},
{
"failure": "Wrong participants (selection bias)",
"symptom": "Interviewing only enthusiasts, or only detractors. Convenience sample (co-workers, friends). Not screening for target criteria.",
"detection": "Check recruitment strategy. Are criteria specific? Is sample diverse? Any obvious biases?",
"fix": "Define precise inclusion/exclusion criteria. Screen with survey. Recruit diverse sample (enthusiasts AND detractors, various demographics). Avoid convenience sampling."
},
{
"failure": "No systematic analysis (cherry-picking)",
"symptom": "No coding or analysis plan. Quotes selected to support pre-existing belief. Contradictory evidence ignored or dismissed.",
"detection": "Check for analysis plan. Is there systematic coding? Are contradictory findings presented? Does report feel one-sided?",
"fix": "Pre-specify analysis approach: thematic coding, affinity mapping, frequency counting. Actively look for disconfirming evidence. Present contradictions. Use inter-rater reliability."
},
{
"failure": "Surface-level probing (stopping too early)",
"symptom": "Accepting first answer without follow-up. Not asking 'Why did that matter?', 'Tell me more', 'What else?'. Missing deeper motivations.",
"detection": "Check interview guide for follow-up probes. Are there '5 whys' style follow-ups? Does guide encourage depth?",
"fix": "Add systematic probes: 'Tell me more', 'Why did that matter?', 'What else?', 'Walk me through what happened next'. Train interviewers to dig deeper."
},
{
"failure": "Feature-focused (not problem-focused)",
"symptom": "Questions about features ('Do you want dark mode?') instead of problems ('Tell me about when poor visibility is an issue'). Solutions before problems.",
"detection": "Count feature mentions vs problem mentions. Are questions about 'what we could build' or 'what problems you face'?",
"fix": "Reframe to problems: 'What challenges do you face with...?', 'When does [current solution] break down?', 'What workarounds have you tried?'. Problems first, solutions later."
},
{
"failure": "No ethics/consent",
"symptom": "Recording without permission. No informed consent. PII not anonymized. No compensation. Participants feel exploited.",
"detection": "Check for consent script, recording permission, anonymization plan, compensation details.",
"fix": "Add consent script. Explicitly ask to record. Anonymize in reports (P1, P2). Offer fair compensation ($50-150 for 60 min). Respect opt-outs."
}
]
}