Subscription Metrics That Matter (And the Ones That Mislead)
MRR, churn, LTV and cohort retention — what each measures, how each is commonly computed wrongly, and which ones actually drive decisions.
A subscription business generates a lot of numbers and most dashboards show the wrong ones prominently.
Here is what each metric is actually for, and how each one lies.
MRR
Monthly recurring revenue — the normalised monthly value of active subscriptions.
What it is for: trend. Is the recurring base growing?
How it lies: it hides composition. MRR flat month-on-month can mean a stable business, or it can mean you lost 200 subscribers and acquired 200 more — which is a very different business with a very different future. Always look at the MRR movement breakdown: new, expansion, contraction, churned, reactivated.
For prepaid-heavy businesses MRR needs care: money received up front is not monthly recurring revenue, and counting it as such overstates the base. Calculating MRR properly.
Churn rate
The proportion of subscribers leaving in a period.
What it is for: the headline health indicator.
How it lies: blended churn mixes cohorts. New subscribers churn much faster than tenured ones, so a business growing fast has churn dominated by new signups. Slow growth, and the same business appears to “improve” — nothing changed except the mix.
Also: a single number hides voluntary versus involuntary, which have entirely different fixes. Churn rate done properly.
Cohort retention
The proportion of a signup cohort still active at cycle 1, 2, 3, and so on.
What it is for: almost everything. It is the honest input to LTV, it shows where you lose people, and it is immune to growth-rate distortion.
How it lies: it does not, much. Its weakness is that recent cohorts have short histories, so you are always extrapolating the latest ones.
If you track one thing, track this. Cohort analysis.
Lifetime value
Expected contribution margin from a subscriber over their life.
What it is for: setting an acquisition ceiling and justifying discounts.
How it lies: in two standard ways. Using revenue instead of contribution margin inflates it by whatever your variable costs are. Using blended churn instead of cohort survival usually inflates it further. Doing it properly.
Renewal success rate
The proportion of billing attempts that succeed.
What it is for: operational health. It is the metric most directly under your control and the one most often ignored.
How it lies: only if you do not segment it by decline reason. A 92% success rate could be 8% insufficient funds (recoverable) or 8% revoked mandates (a churn signal). Same number, opposite implications. Failed payments.
Metrics that mislead more than they help
Total subscribers. Grows even as the business deteriorates, if acquisition outpaces churn. Says nothing about quality.
Average order value on subscriptions. Moves when your plan mix shifts, which is usually not what you were trying to measure.
Conversion rate to subscription without retention context. Pre-selecting the subscribe option lifts this number and can lower revenue, because the subscribers it adds are the ones who did not mean to subscribe.
The short list
If you track four things:
- Cohort retention by cycle — where you lose people
- MRR movement — new, expansion, contraction, churned
- Renewal success rate by decline reason — the operational lever
- Contribution margin per cycle — whether any of it is profitable
Everything else is a derivative of those.
Frequently asked questions
What is the most important subscription metric?
Cohort retention by cycle number. It is the input to lifetime value, it exposes whether early or late churn is your problem, and unlike blended churn it does not move when your growth rate changes.