Stop guessing. Build a repeatable system that reduces bias and turns
user needs into measurable outcomes.
Why Teams Fail at Understanding Users
Many product teams believe they deeply understand their customers. In
reality, they often project their own assumptions.
Psychological research on empathic accuracy shows that humans
systematically overestimate their ability to infer others' internal
states. Intuition feels accurate --- but frequently isn't.
That's why discovering user needs must be systematic, not emotional.
Core Principles
1. Design for Outcomes, Not Desires
Instead of asking what users want, identify what they are trying to
achieve.\
Turn needs into measurable outcome statements.
Example:
When buying on mobile at night, users want to complete checkout in
under 3 minutes so they avoid friction and abandonment.
If it cannot be measured, it cannot be validated.
2. Use Evidence Triangulation
Never rely on one source.
Combine: - Behavioral analytics (what users do) - Interviews (why they
do it) - User-generated content (reviews, tickets, forums) - Controlled
experiments
When multiple signals converge --- confidence increases.
3. Interview for Behavior, Not Opinion
Bad question: > "Would you use this feature?"
Good question: > "Tell me about the last time you tried to solve this
problem."
Focus on: - Recent real events - Step-by-step actions - Workarounds -
Frequency - Emotional friction points
Avoid hypotheticals. Humans predict behavior poorly.
Practical Workflow (6 Steps)
Step 1 --- Define a Precise Question
Turn curiosity into a testable hypothesis.
Bad: > "How can we improve onboarding?"
Good: > "What is the largest friction point preventing first-time users
from completing onboarding within 5 minutes?"
Step 2 --- Run Structured Interviews
Use semi-structured interviews.
Template: - What triggered the action? - What alternatives did you
consider? - What made it difficult? - How often does this happen? - What
metric defines success?
Step 3 --- Translate Insights into Outcome Statements
Template:
When [situation], the user wants to [functional job], measured by
[metric], so they can [desired higher-level outcome].
Example:
When exporting reports, the user wants export time under 10 seconds so
workflow is not interrupted.
Step 4 --- Instrument Metrics
Track the exact metrics named in outcome statements.
Examples: - Time to complete task - Error rate per session - Drop-off
percentage - Retry frequency
Step 5 --- Validate with Experiments
Ship small changes. Measure impact. Rollback if metrics do not move.
Evidence over ego.
Step 6 --- Maintain a Research Log
For every product decision document: - Interview evidence - Quantitative
data - Experiment results - Outcome metric movement
Make decisions auditable.
Common Illusions to Avoid
- "We just know our users."\
- "One interview confirmed it."\
- "The feature feels right."
Feelings are hypotheses --- not validation.
Final Checklist Before Shipping
- Is the need measurable?
- Do we have at least 2 evidence sources?
- Is there an experiment designed?
- Are we ready to admit we were wrong?
Closing Thought
Smart teams don't guess better.\
They measure better.
Build systems that reveal truth --- not systems that protect
assumptions.