Separate cancellation reasons
A merchant who never finished setup is a different problem from one who used the app successfully for a year and then closed the store. Keep the reason list short: setup failure, missing capability, insufficient value, technical issue, price, business closure, unknown. Attach a note with the specific circumstances.
Do not fill the unknowns with whichever reason you find most comfortable. A cancellation with no feedback attached is unknown, and it stays unknown until evidence arrives.
Review revenue as a movement, not a total
Break the month into the pieces that moved MRR.
| Movement | MRR |
|---|---|
| Starting MRR | $5,000 |
| New customers | plus $1,500 |
| Existing customers paying more | plus $200 |
| Downgrades | minus $100 |
| Cancellations | minus $600 |
| Ending MRR | $6,000 |
Net new MRR is $1,000. Gross new customer MRR is $1,500. They answer different questions: the first says how the business moved, the second says how well acquisition performed.
Inspect churn with the right denominator
Customer count and revenue can tell different stories.
You start a month with 50 paying customers and $5,000 MRR. Five cancel, taking $800 of MRR with them. Customer churn is 5 / 50 = 10%. Cancellation MRR churn is $800 / $5,000 = 16%. The revenue loss is larger because the accounts that left paid more than average, and that should change your diagnosis. A customer-count report alone would have hidden the concentration.
Now suppose the remaining customers add $300 of MRR by moving up plans. Net revenue retention for the starting group, excluding new customers, is ($5,000 minus $800 plus $300) / $5,000 = 90%, assuming nothing else moved. State those inclusions whenever you use the metric.
Track the first renewal separately
The first renewal tells you something the first payment cannot. A merchant will often pay during setup and then stop when the app never becomes part of how they work. For monthly subscriptions, track what share of each new cohort is still paying after its first renewal opportunity and the ones after that.
For annual subscriptions, early usage and customer outcomes give interim evidence, but the renewal stays unobserved until the contract period ends. Do not compare annual and monthly retention as though the observation opportunities were equivalent. A customer nine months into an annual contract can be commercially active and operationally disengaged at the same time.
Use cohort retention before broad averages
Compare acquisition sources and segments with similar time at risk. A recent cohort has had fewer opportunities to cancel than an older one, and comparing them directly flatters the newer source. Customers acquired through compensated meetings may behave differently from those who arrived through category search, with different motivation and setup needs, so compare their activation and retention before deciding which source produces better customers.
Keep the sample size visible. Two retained accounts are experience, not a retention rate you can forecast a budget against. Retention also starts with the customer you select: keep the original promise and merchant profile attached to the account, and compare both with what was delivered when someone cancels.
Keep an account outcome record
Record the merchant's desired result, the activation date, current product use where you can observe it, support issues and the next review. A self-serve app can automate much of that, while an assisted account may need a named owner and an outcome conversation. Do not report activity as value. Logging in is evidence of access, not of completing the workflow that was supposed to help.
Fix problems before you ask for anything. Support must never depend on a favorable review or a referral. Where you run a customer referral incentive, disclose the terms and keep it separate from review requests. A referral reward is a commercial program, and an honest product review is something else.
Is acquisition or retention the constraint?
Suppose you add $1,500 of gross new MRR a month and lose $1,200. Doubling outbound increases net growth and also feeds a leaky system. First work out whether the losses come from fit, activation, product value or ordinary business turnover, because only the first three are yours to fix. Some loss is unavoidable. The question is whether an identifiable retention improvement is a better use of your time and cash than another acquisition campaign.
Do this week
- Define who owns activation, and write your short list of cancellation reasons.
- Compare your last five cancellations against five retained customers from similar segments.
- Pick one preventable cause and track the next comparable cohort.
- Report customer churn, MRR churn, net new MRR and first-renewal retention by acquisition source, with sample sizes.
Questions
What is net new MRR for a Shopify app?
It is ending MRR minus starting MRR after all movements: new customers, expansion, downgrades and cancellations. Gross new customer MRR counts only the new customers, so the two measure different things. With $5,000 starting, plus $1,500, plus $200, minus $100, minus $600, net new MRR is $1,000.
Why is MRR churn different from customer churn?
Customer churn counts accounts and MRR churn counts revenue. If the customers who leave paid more than average, MRR churn is higher: 5 of 50 customers is 10%, but $800 of $5,000 MRR is 16%.
How do I measure retention on annual plans?
Use early usage and customer outcomes as interim evidence, and treat the renewal as unobserved until the contract ends. Do not compare annual and monthly retention as though the observation opportunities were equivalent.
Related guides
Chapter 7: How to turn Shopify app installs into paying customersChapter 18: How to build a 90-day go-to-market plan for a Shopify appNumbers marked as planning examples are arithmetic illustrations, not Shopify benchmarks or promised results. Founder-reported figures are not independently audited. Platform rules change: confirm current requirements with Shopify before you act.