How would you diagnose a flat feature adoption KPI?

What's really being asked

This question assesses your product sense and data literacy. The interviewer wants to see if you can think systematically, moving from a high-level, lagging indicator (adoption is flat) to a concrete, actionable plan. They are testing your ability to formulate hypotheses, define a user funnel, identify relevant leading indicators, and combine different types of data (quantitative and qualitative) to find the root cause. It's not about knowing one right answer, but about demonstrating a structured diagnostic process.

The full answer

A strong answer outlines a multi-stage diagnostic plan. First, define the user funnel for the feature: Awareness (did users see it?), Activation (did they try it once?), and Usage (did they use it correctly and repeatedly?). Second, for each stage, propose specific metrics and tools. For Awareness, investigate discovery paths, click-through rates on banners, or heatmap analysis. For Activation, analyze funnel drop-off rates during the feature's setup or first use. For Usage, look at frequency and repeat engagement. Third, stress the need to combine this quantitative data (the "what") with qualitative data (the "why") using tools like session replays to see where users struggle, or targeted in-app surveys to ask them directly. Finally, mention segmenting the data by user cohorts (new vs. existing, power users vs. casuals) to find hidden patterns.

The mistakes people make

A major red flag is jumping to a conclusion or a single solution without a diagnostic framework. Saying "We probably have an onboarding problem, so I'd rebuild the tutorial" is a weak answer because it assumes the problem without evidence. Another poor response is listing generic metrics without tying them to a specific stage of the user funnel or a hypothesis. The goal is to show a process for finding the problem, not to guess the solution. Blaming external factors without looking at internal data first is also a red flag.

What usually comes next

Expect questions like: "Let's say you find the drop-off is at the Activation stage. What are three specific hypotheses you'd test?" or "How would you prioritize which of these analytics to implement first, given limited engineering resources?" or "What if you have the quantitative data but no qualitative tools? How would you get the 'why'?"

A concrete example

If our feature is a new "Advanced Search Filter" and adoption is flat at 2%, my plan is: 1) Awareness: Is the filter visible? I'll check heatmaps and measure the percentage of users who even see the filter control. 2) Activation: Of those who see it, how many click to open the options? I'll use funnel analysis to find the drop-off; maybe the UI is confusing. 3) Qualitative: I'll watch session replays of users who hover over the filter but don't use it to understand their hesitation.