Most product teams know how to measure growth: you track new customers, increase usage, and improve average transaction value. But what about churn — the customers you lose? That’s harder to pin down, especially for products without predictable usage patterns. Most product teams know how to measure growth: you track new customers, increase usage, and improve average transaction value. But what about churn — the customers you lose? That’s harder to pin down, especially for products without predictable usage patterns. Unlike signup or usage metrics, churn isn’t always visible. A customer who hasn’t used your app in weeks might have left… or they might just be busy. This ambiguity makes it difficult to measure churn, prioritize work on it, or tie improvements to business impact. Unlike signup or usage metrics, churn isn’t always visible. A customer who hasn’t used your app in weeks might have left… or they might just be busy. This ambiguity makes it difficult to measure churn, prioritize work on it, or tie improvements to business impact. In this article, I’ll share a hands-on framework that makes churn more measurable and actionable — even for irregular-use products — by borrowing from conversion funnel analysis and applying it to retention. In this article, I’ll share a hands-on framework that makes churn more measurable and actionable — even for irregular-use products — by borrowing from conversion funnel analysis and applying it to retention. The Problem with Measuring Churn The Problem with Measuring Churn For subscription-based services or regularly used products, churn is easy to identify. If a customer doesn’t pay or misses their scheduled interaction, they’ve likely churned. But many digital products — especially transactional ones — don’t work that way. Use patterns are sporadic and vary by user. For example, someone might use a food delivery app daily for two months, then stop completely. Did they churn? Or did their habits just change? To compensate, teams often define a fixed “inactivity window” — say, 30 days — after which a user is considered churned. But this creates two problems: False positives: Some users marked as churned may come back later. Poor fit across verticals: Thirty days might work for games or social apps, but it’s too short for products like international money transfers or e-commerce. False positives: Some users marked as churned may come back later. Poor fit across verticals: Thirty days might work for games or social apps, but it’s too short for products like international money transfers or e-commerce. So, how do we define churn more accurately? A Better Approach: Conversion to the Next Transaction A Better Approach: Conversion to the Next Transaction Think of churn not as an abstract disappearance, but as a failure to convert to the next transaction. Just like we track conversion funnels for acquisition (visit → signup → first transaction), we can track similar funnels post-purchase. The idea: after a customer completes a transaction, what are the odds they’ll return and do another one? Step 1: Build the Curve Step 1: Build the Curve Start by taking a large sample of transactions — say, 1,000 — and plot the following: X-axis: Days since transaction Y-axis: Percentage of those transactions that led to a repeat transaction within X days X-axis: Days since transaction Y-axis: Percentage of those transactions that led to a repeat transaction within X days The resulting chart will show how many customers come back — and how quickly. Some will transact again the same day, others after a week or two. Eventually, the curve flattens out: no more customers are converting. This “asymptote” is your churn baseline. For example, if your curve flattens at 90%, that means 10% of those customers never return. That’s your true churn rate. Now, let’s say you handle 1,000 transactions per month. That 10% churn equals 100 customers lost monthly. Suddenly, churn isn’t a vague concept — it’s a number you can tie to revenue and retention strategy. Estimating Retention Gains Estimating Retention Gains This approach also lets you forecast the impact of improvements. If you find a way to raise that asymptote from 90% to 93%, you’ve retained 3% more users — or 30 more customers per month. You can now compare that impact directly to growth initiatives like adding 30 new users via acquisition. Better yet, you can use this chart to define a smarter inactivity window. Let’s say after 18 days, 80% of users have returned, but 10% more trickle in after that. If you label users inactive at 18 days, you’re getting a 50% false-positive rate. This gives you an informed way to balance detection speed against prediction accuracy. From Detection to Diagnosis: Survey the “Churned” From Detection to Diagnosis: Survey the “Churned” Once you’ve identified users who didn’t convert within your selected time frame, the next step is to ask why. Here’s how. Survey users marked as churned with a simple question: “We noticed you haven’t used the product recently — can you tell us why?” Provide response options such as: “I started using another product instead” “I don’t need this product anymore” “I plan to use it again soon” “I started using another product instead” “I don’t need this product anymore” “I plan to use it again soon” This helps distinguish between: Switchers: Actionable churn (they chose another product) Lost use case: Inherent churn (they no longer need your solution) False positives: Still engaged (they plan to return soon) Switchers: Actionable churn (they chose another product) Lost use case: Inherent churn (they no longer need your solution) False positives: Still engaged (they plan to return soon) For switchers, follow up with: What product did you switch to? Why? (cheaper, faster, more features, etc.) What product did you switch to? Why? (cheaper, faster, more features, etc.) This provides both quantitative churn reasons and competitive intelligence. You’ll learn what’s driving attrition — and whether it’s a pricing issue, a UX flaw, or missing functionality. Estimating Churn That’s Fixable Estimating Churn That’s Fixable Let’s say 25% of respondents say they switched products. If you have 1,000 users per month and estimate 10% churn (100 users), then 25 of them are leaving for a competitor. That number — 25/month — becomes your actionable churn. You can now measure and prioritize projects against it: Would improving onboarding reduce this number? Would a missing feature win them back? Could pricing adjustments move the needle? Would improving onboarding reduce this number? Would a missing feature win them back? Could pricing adjustments move the needle? Add this “monthly switched users” KPI to your roadmap planning. It gives you a tangible, user-driven input for decision-making. Final Thoughts Final Thoughts Churn doesn’t have to be a black box. By looking at conversion to the next transaction, you turn it into a measurable, analyzable event. You can identify true churn, predict it with more accuracy, and survey users to uncover root causes. Best of all, this method helps you prioritize retention work. You can size the impact of improvements, compare churn reduction to acquisition gains, and track progress with real numbers. If you’ve struggled with retention analysis — or just want to dig deeper than LTV or NPS — try this approach. It might change the way your team thinks about user growth.
