How to answer retention metric questions for subscription products
Before recommending retention improvements for a subscription product, clarify what counts as churn, segment users by behavior, and target changes based on root causes. Use a step-by-step example to show how your assumptions affect metrics, then propose a realistic way to test your thinking.
PMMockr.com
Clarify the churn definition before calculating retention
A common mistake in retention metric questions is jumping straight into solutions without defining churn. In subscription products, churn can mean voluntary cancellations, failed payments, or even users who simply stop using the service while still paying. Each of these scenarios affects retention differently and points to different interventions.
Start by asking how churn is measured: Is it based on billing cycle non-renewals, account inactivity, or failed payment attempts? This ensures your answer fits the product's business logic and avoids mismatched recommendations. For example, a streaming service may consider anyone who stops paying as churned, while a productivity tool might look at active usage in addition to payments.
Why segmentation matters for interpreting retention metrics
Retention is rarely uniform across a subscription base. Segmenting users by tenure, acquisition channel, plan type, or engagement level can expose patterns that aggregate metrics hide. For instance, users who signed up for a discounted annual plan may behave very differently from monthly trial users.
Segmentation helps prioritize where to intervene. If new users churn at a high rate but long-term users are stable, onboarding deserves focus. If annual subscribers unexpectedly cancel mid-term, it may signal dissatisfaction or a competitive threat. Always ask for or propose at least one meaningful segment in your response.
Retention is a curve, not a single number
Interviewers may ask for a retention rate, but experienced PMs know retention unfolds over time. A 30-day retention figure differs in meaning from a 12-month one. For subscription products, a retention curve shows the percentage of users remaining subscribed at each time point after signup.
Discussing the retention curve demonstrates that you recognize onboarding, habit formation, and long-term value are distinct challenges. It also sets you up to recommend interventions that target the steepest drop-offs, not just the overall average.
Worked example: Improving 90-day retention for a meditation app
Assume you are given this scenario: A meditation app with a monthly subscription has a 90-day retention rate of 60%. Of 1000 new subscribers, 600 are still paying after three months. Management wants to improve this metric.
Start by clarifying churn. Ask: Does this 60% include only voluntary cancellations, or does it also count users lost to failed payments? Assume for this exercise that 10% of churn is due to payment failure (involuntary churn), while 30% is voluntary cancellation.
Segment users by signup channel. Suppose 700 of the 1000 users came via a free 7-day trial, and 300 paid upfront. After 90 days, only 50% of trial users remain versus 80% of upfront payers. This suggests the free trial cohort is driving most churn.
Given this, recommend focusing on trial-to-paid conversion and early engagement. For example, propose an onboarding flow that guides trial users to set reminders for their first week and offers a personalized meditation plan. This targets the segment with the steepest retention curve drop-off. As a competing option, you could suggest stricter payment retries to recover involuntary churn, but this only addresses 10% of lost users.
The trade-off is between fixing a small but easy problem (payment failures) and tackling the larger, trickier issue of voluntary cancellation. Prioritize the latter for greater impact. To validate, run an A/B test on the new onboarding flow for trial users and measure its effect on 90-day retention. If the improvement is less than expected, revisit your assumption that onboarding is the main driver—perhaps content relevance or pricing is the real issue.
Making assumptions explicit builds credibility
Strong retention answers state assumptions clearly, allowing the interviewer to correct or refine them. In the meditation app example, you specify the breakdown between involuntary and voluntary churn, and the signup channels’ retention differences.
If you simply assert 'trial users churn more' without data or explanation, it sounds like guesswork. By labeling invented inputs and showing your reasoning, you invite constructive feedback and signal analytical rigor.
Comparing weak and strong reasoning
A weak response to a retention metric question might be: 'We should add a loyalty program to reduce churn.' This skips over the size of the problem, ignores user segments, and doesn’t explain why a loyalty program would work.
A stronger alternative: 'Since 70% of churn comes from trial users who don't convert after 7 days, and our data shows they are less engaged in the first week, I recommend targeted onboarding changes. This addresses the largest group at risk and is measurable via a controlled experiment.' The difference is the focus on evidence, segmentation, and testable actions.
Choosing metrics for retention experiments
When proposing retention improvements, be specific about which metric will signal success. For the meditation app, the primary metric is 90-day paid retention for new trial users. Secondary metrics might include week-one engagement or conversion from trial to paid.
Make sure your experiment design matches the metric. For instance, onboarding changes should track whether more users remain subscribed after 90 days—not just short-term engagement spikes. This prevents premature conclusions based on vanity metrics.
Checking your decision: What would change your recommendation?
After implementing the onboarding experiment, suppose retention among trial users rises from 50% to 55%—a modest improvement. If this falls short of expectations, consider what evidence would shift your approach. For example, qualitative feedback might reveal users cancel due to limited content variety, not onboarding friction.
In this case, you might deprioritize onboarding tweaks and instead propose expanding the meditation library or tailoring content recommendations. By defining what new data would change your mind, you show adaptability and a hypothesis-driven mindset.
Practice exercise and self-review checklist
Timed exercise (15 minutes): Given a fictional streaming service with a 6-month retention rate of 40%, draft a response outlining how you would investigate and improve this metric. Specify your assumptions about churn, propose a user segment to focus on, and describe a concrete experiment. State what result would lead you to change your recommendation.
Self-review checklist: - Did I clarify how churn is defined? - Did I segment users meaningfully? - Did I propose a specific and measurable change? - Did I make assumptions explicit? - Did I state what evidence would shift my approach?
Practice this style of reasoning twice a week with new hypothetical products or metrics to build fluency. PMMockr offers additional practice questions to reinforce these habits.
FAQ
How do I handle ambiguous churn definitions in interviews?
Ask the interviewer to clarify what counts as churn—voluntary cancellations, payment failures, or inactivity. If that's not possible, state your assumption explicitly before continuing your analysis.
Should I always segment by acquisition channel?
Segmenting by acquisition channel is often useful, especially when different channels bring users with distinct behaviors. If other segments (such as plan type or engagement level) are more relevant, prioritize those instead.
What if my retention experiment doesn't work as expected?
Explain how you would interpret the results and what alternative hypotheses you would consider. Show that you are prepared to adapt your recommendation based on new evidence, not just stick with your original plan.