Combine qualitative and quantitative data for hypotheses

WHAT THIS TESTS This evaluates whether you understand that quantitative and qualitative data answer different questions and are strongest together. The interviewer wants to see a disciplined path from observation to a falsifiable hypothesis, not a leap from a single user quote to a feature.

A GOOD ANSWER COVERS Start with quantitative analytics to find where and how large a problem is. Funnel analysis, cohort retention, and segment breakdowns show that, for example, users drop sharply at a particular onboarding step. Quant tells you what is happening and at what scale but rarely why. Then use qualitative methods, user interviews, session replays, and support tickets, to uncover the mental model behind the behavior: maybe users do not understand what a step asks for. Triangulate by checking whether the interview-derived mechanism matches the quantitative pattern across segments. If several users describe confusion and the data shows the drop concentrated among new users, the signals reinforce each other. Finally, write a falsifiable hypothesis: if we clarify the step's copy, then new-user completion of that step will rise, measured by step completion rate. Specify the metric, the expected direction, and the segment.

COMMON WRONG ANSWERS Treating a few vivid interviews as statistically representative. Staring at dashboards and inventing a why with no user contact. Producing a vague hypothesis with no metric or direction. Cherry-picking quotes that confirm a preexisting belief. Confusing correlation in analytics with the cause users actually report.

LIKELY FOLLOW-UPS How many interviews are enough to form a hypothesis? Enough to hear themes repeat, not to prove prevalence. How do you handle conflicting signals? Investigate the gap rather than pick the convenient one. How do you avoid leading questions in interviews?

ONE CONCRETE EXAMPLE Analytics show a forty percent drop at the address-entry step of checkout, worst on mobile. Five interviews reveal users abandon because the form rejects valid formats without explaining why. The hypothesis becomes: adding inline format hints and lenient parsing will lift mobile checkout completion at this step, measured over a two-week A/B test on the step completion rate.