How would you estimate user base for a new feature before coding?

WHAT THIS TESTS: This question tests whether you can de-risk product bets with data before committing engineering time. Interviewers want to see that you understand how to construct proxy audiences, triangulate multiple data sources, and communicate uncertainty. It is not about getting the exact number right; it is about demonstrating structured thinking and business pragmatism.

A GOOD ANSWER COVERS: First, define the behavioral profile and the job-to-be-done so you know what signal to proxy. Second, identify internal analogs such as adoption curves from similar features or adjacent user segments already in your data warehouse. Third, layer external proxies including competitor public metrics, industry reports, and demand signals like search volume or social listening. Fourth, synthesize these into a bounded range using a top-down TAM sanity check and a bottom-up funnel with assumed conversion rates at each step. Fifth, explicitly state your assumptions and confidence intervals so stakeholders understand the error bars.

COMMON WRONG ANSWERS: A major red flag is insisting that no estimate is possible without shipping an MVP first. Another weak pattern is relying solely on gut instinct or executive opinion without any data anchor. Giving a single point estimate without ranges or assumptions is also problematic because it hides uncertainty. Finally, using only one data source, such as just a survey or just a competitor press release, shows a lack of triangulation discipline.

LIKELY FOLLOW-UPS: The interviewer may ask how you would validate the proxy after launch, how you would prioritize this feature against others with firmer data, or what you would do if internal analogs do not exist. They might also probe how you would adjust the estimate if the feature targets a niche B2B audience rather than a broad consumer market.

ONE CONCRETE EXAMPLE: Suppose you are launching a hard seltzer brand and no direct consumer data exists yet. Following the proxy audience framework, you would first build a lifestyle audience around demographics and psychographics of health-conscious younger drinkers. If the segment is niche, you would layer in publications and influencers they follow. If you are an established brewery looking at growth, you would start with competitor seltzer audiences and refine them with your existing customer demographics and consumption data. You might then cross-reference this against search trend volume and liquor-store category growth reports to bound the market size before producing a single can.

Source: telmarhelixa.com