You have 1,000 Stripe subscribers. How many will cancel next quarter — and which ones? Churn prediction analyzes your subscription data to measure churn rates by plan, identify at-risk accounts, quantify the MRR you stand to lose, and show you exactly where to focus retention efforts. Export your Stripe subscriptions, upload the CSV, and get answers in under 60 seconds.
What Is Churn Prediction?
Churn prediction answers the question every subscription business asks: which customers are about to leave, and why? It takes your Stripe subscription data — plan types, billing intervals, subscription ages, cancellation history — and uses survival analysis techniques to estimate the probability that each subscriber will cancel within a given time window.
Think of it this way. You run a SaaS product with three tiers: Basic at $29/month, Pro at $79/month, and Enterprise at $199/month. You have 1,000 active subscribers. Churn prediction does not just tell you that your overall churn rate is 5% per month. It tells you that Basic monthly subscribers churn at 8.2%, Pro annual subscribers churn at 1.1%, and that subscribers who survive their first 90 days are three times more likely to stay for a year. That specificity is what turns a churn number from a depressing metric into an actionable retention plan.
The analysis works by examining the lifecycle of every subscription in your dataset — when it was created, what plan it is on, how it is billed, whether a trial was involved, and whether it has been canceled or is flagged for cancellation at period end. Subscriptions that are still active are treated as "censored" observations, meaning we know they have survived at least this long, but we do not yet know when (or if) they will cancel. This is the same survival analysis framework used in clinical trials and insurance actuarial work, applied to your revenue.
When to Use Churn Prediction
The primary use case is any business that bills customers on a recurring basis through Stripe. SaaS companies are the obvious fit — you want to know which accounts need a check-in call before they hit the cancel button. But subscription boxes, membership businesses, digital media platforms, and telecom providers all face the same problem: recurring revenue only recurs if customers stay.
Use this analysis when you want to answer specific operational questions. Which plan tier has the worst retention? Is monthly billing driving more churn than annual? Do customers who start with a trial convert and stick, or do they churn faster than customers who paid from day one? How much MRR is at risk right now from subscriptions that have been flagged for cancellation at period end?
Timing matters. Run churn prediction quarterly to track trends and update your at-risk list. Run it after a pricing change to see if the new structure is affecting retention. Run it before budgeting season to forecast expected revenue loss and justify retention investment. A subscription business that does not measure churn by segment is flying blind — aggregate churn rates hide the segments that are hemorrhaging and the segments that are rock solid.
This analysis is also valuable for investor reporting. Gross churn rate, net revenue retention, and MRR at risk are metrics that every SaaS investor scrutinizes. Having a data-backed breakdown by plan tier, with survival curves and hazard ratios, gives your board deck substance instead of a single top-line number.
What Data Do You Need?
You need a CSV export of your Stripe subscriptions. The analysis requires five columns and supports six optional columns that unlock deeper insights.
Required columns:
- sub_status — The current subscription status (active, canceled, past_due, trialing, etc.). This is how the analysis determines which subscriptions have churned and which are still alive.
- plan_name — The name of the subscription plan (Basic, Pro, Enterprise, or whatever your tiers are called). This drives the plan-level segmentation that makes the analysis actionable.
- plan_amount — The price of the plan in your currency. Used to calculate MRR, revenue at risk, and the financial impact of churn by tier.
- plan_interval — The billing frequency (month, year, week). Monthly vs. annual billing often has dramatically different churn characteristics, and this column lets the analysis separate them.
- created_date — When the subscription was created. This anchors the survival analysis timeline and enables subscription age calculations.
Optional columns (recommended):
- canceled_date — When the subscription was actually canceled. Without this, the analysis infers cancellation from status alone, which is less precise.
- cancel_scheduled — Whether cancellation is scheduled at period end (cancel_at_period_end in Stripe). This is a critical early warning signal — these subscriptions are walking out the door but have not left yet.
- trial_start and trial_end — Trial period dates. Enables the trial conversion analysis card, which shows how many trial users convert to paid and whether trial-origin subscribers churn differently.
- seat_quantity — Number of seats or licenses. Useful for per-seat SaaS products to weight churn by account size.
- billing_method — How the subscription is billed (charge_automatically, send_invoice). Invoice-billed accounts sometimes have different churn patterns due to payment friction.
For reliable results, you need at least 100 subscriptions. The analysis works best with 500 or more subscriptions and at least six months of history, so the survival curves have enough data to stabilize. If you have fewer than 100 subscriptions, the churn rates will be noisy and the plan-level breakdowns may not have enough observations per segment to be meaningful.
How to Read the Report
The report opens with the Churn Executive Summary — your top-line numbers. Overall churn rate, total MRR, MRR at risk, and the number of at-risk subscriptions. This is the slide for your board meeting. If only one number matters, it is the MRR at risk figure, because it translates churn from a percentage into dollars you are about to lose.
The Subscription Status Distribution card shows the breakdown of your subscriber base by status — active, canceled, past_due, trialing, and any other Stripe statuses in your data. A healthy business has a large active segment and a small canceled segment. If past_due is growing, you have a payment recovery problem that is separate from voluntary churn.
The MRR by Plan and Status card is where the financial picture comes into focus. It cross-tabulates your monthly recurring revenue by plan tier and subscription status. You might discover that your Enterprise tier contributes 60% of MRR but only 15% of churn, while your Basic tier generates 10% of MRR but accounts for half of all cancellations. That tells you where retention investment has the highest return per dollar.
The Churn Risk by Plan Tier card segments churn rates by plan. This is the card that usually produces the biggest surprises. You may assume all plans churn equally, but the data almost always shows significant variation. A SaaS company might see 12% monthly churn on their Basic plan and 2% on Enterprise — a six-fold difference that demands different retention strategies for each tier.
The Subscription Age Distribution card shows how long your subscribers have been active. Survival analysis tells us that churn risk is not constant over time — it is usually highest in the first 30 to 90 days and drops sharply after that. If you see a spike of cancellations in month two, your onboarding is failing. If churn is steady across all ages, you have a product-market fit problem rather than an onboarding problem.
The At-Risk Subscriptions card lists individual subscriptions that are flagged for cancellation or are in a past_due state. These are the accounts you can still save. The list includes plan name, amount, and how long they have been subscribed — giving your customer success team a prioritized outreach list sorted by revenue at stake.
If you included trial data, the Trial Conversion Rates card shows what percentage of trial users convert to paid subscribers and how their subsequent retention compares to subscribers who never had a trial. Some businesses find that trial users convert well but churn faster; others find that trials filter out poor-fit customers, leading to better long-term retention.
When to Use Something Else
If you want a more detailed survival analysis with hazard ratios that quantify exactly how much each factor (plan type, billing interval, trial status) increases or decreases churn risk, use Cox Proportional Hazards. Cox regression gives you coefficients you can interpret as risk multipliers — for example, "monthly billing has a 2.3x higher hazard of churn compared to annual billing, controlling for plan tier." This analysis gives you the segmented churn rates; Cox gives you the multivariate model.
If your question is simpler — you just want a binary prediction of "will this customer churn yes or no" based on a set of features — logistic regression is a good fit. It does not model the time dimension (when they will churn), but it gives you a probability score you can use for lead scoring and outreach prioritization.
If you want to understand customer behavior patterns rather than predict cancellation, consider RFM segmentation. RFM groups customers by Recency, Frequency, and Monetary value, which is complementary to churn prediction — it tells you who your best customers are, while churn prediction tells you who is about to stop being a customer.
If you are looking at retention from a cohort perspective — what percentage of customers who signed up in January are still active in March — use cohort retention analysis. Cohort analysis shows retention curves over time for groups defined by signup date, which is a different lens than the plan-based segmentation this tool provides. The two analyses are highly complementary.
The R Code Behind the Analysis
Every report includes the exact R code used to produce the results — reproducible, auditable, and citable. This is not AI-generated code that changes every run. The same data produces the same analysis every time.
The analysis uses the survival package for survival curve estimation via survfit() and Kaplan-Meier curves that show the probability of a subscription surviving past each time point. Churn rates are calculated per plan tier using status-based classification, and MRR at risk is computed by joining plan amounts with at-risk status flags. The dplyr and lubridate packages handle the date arithmetic for subscription age calculations and cohort bucketing. Every step is visible in the code tab of your report, so you or your data team can verify exactly what was done and reproduce the results independently.