Summary
A workshop-style “User Research Fellowship” module (Nov 2022) outlining a pragmatic research practice—from problem framing to sampling, methods, bias management, tooling, and planning. The throughline is rigor without rigidity: adapt methods to objectives, build redundancy in inputs, and design safeguards against bias and logistics failure.
What this covers
- Problem identification & framing (with guiding principles + mapping exercises)
- Sourcing inputs & defining context (stakeholders, influence, and research redundancy)
- Hypothesis building + sacrificial concepts (learn fast through disposable prompts)
- Methods overview (what to keep sacred, what to adapt)
- Table-stakes planning (effort, logistics, contingencies, roles)
- Cultural context & bias management (with emphasis on India)
- Sampling (representative and realistic)
- Tooling (research, documentation, synthesis)
- Tips, traps, watch-outs (participant dynamics, fatigue, scheduling, debriefs)
Key frameworks & ideas
1) Problem framing principles
Move from:
- Business → Human → behavioural/emotional
- Solution-first → problem-first
- Generic → nuanced/contextual
- Surface-level → underlying/fundamental
- Negative → positive (where possible)
- Open-ended → directional
2) Context + input redundancy
Use multiple sources intentionally (past research, stakeholder interviews, audits, quant sources, analogous inspiration, credible frameworks). Define the objective before “going hunting.”
3) Hypotheses + sacrificial concepts
- Descriptive hypotheses explain behaviour.
- Prescriptive hypotheses aim to influence behaviour.
- Sacrificial concepts are intentionally rough, unique, and discardable—used to provoke conversation and reveal assumptions early.
4) Cultural context + bias acknowledgement
Bias can enter via the researcher, intent, tools, participants, and context. Common traps: confirmatory framing, leading/close-ended questions, over-indexing opinions/rationalisations, and “lies honestly told” in hot states.
5) Sampling (often overlooked)
Avoid self-selection and purely demographic segments; prioritise behavioural segments, extreme users, and positive deviances. Rule of thumb: qual 5–7 per type (until saturation), quant 50+; plan backups and screen carefully.
Key takeaways
- Frame the human behavioural problem before designing a solution.
- Build redundancy in inputs, but stay objective-led.
- Use sacrificial concepts to surface what you don’t know—fast.
- Treat sampling as a design problem, not a recruiting task.
- Bias management begins with acknowledgement and guardrails.
Credits
User Research Fellowship | Shivani Gupta | Nov 2022