Your Users Are Lying to You

You ask a user, “Would you use this feature?” They say yes. You build it. They never open it.

The title says they lied. Usually they did not. You asked them to predict an imaginary future, and they gave you a polite intention.

Do Not Ask Whether Someone Would Join the Gym

Many people honestly intend to exercise next month. That answer does not tell you whether they will wake up early, travel to the gym, and repeat the behavior after the first week.

A feature question has the same weakness. Saying yes costs nothing. Using the feature requires the right problem, moment, effort, and trade-off to appear in real life.

Intentions are useful context. They are weak evidence of behavior.

Ask About the Last Real Situation

Replace “Would you use this?” with “Tell me about the last time this happened.” Then stay with the sequence.

• What triggered the situation?
• What did the person do first?
• Which workaround or tool did they use?
• What did the problem cost in time, money, delay, or risk?
• Why did they stop, continue, or choose another route?

A person who manually combines three reports every Monday gives you a behavior to inspect. A person who says a unified dashboard “sounds useful” gives you an opinion.

Look for Commitment at the Right Scale

Payment is strong evidence, but it is not the only evidence. Time invested, data shared, a pilot scheduled, a workflow changed, or repeated use can all show commitment. The required signal should match the stage and cost of the decision.

Do not demand a purchase before a problem can be explored. Also do not treat enthusiasm as proof that justifies months of development.

Usage Data Needs Interpretation Too

Behavioral data is closer to reality, but it does not explain itself. If users abandon a flow, they may not need the feature, or the flow may be confusing, broken, slow, or asking for trust too early.

Combine observation with conversation. Ask users what happened around the behavior without asking them to invent a clean explanation for your analytics.

Collect Evidence, Not Agreement

Your research has failed if every conversation merely confirms the original idea. The job is to discover which assumption is weak while the team can still change it cheaply.

Users do not owe you a product decision. Give them questions they can answer from life, then make the decision from the evidence.