What subscription revenue leakage actually means
Revenue leakage is not a single number. It is a class of gaps between what a customer was billed and what the product actually allowed. This piece defines the categories and the evidence needed to investigate them.
Subscription revenue leakage usually gets described as “lost MRR.” That framing is too narrow. A leak can be a paying customer who cannot use the product, a former customer who still can, a plan mismatch that gives away an expensive feature, or a billing record that cannot be tied to any account at all. Some of these cost money immediately. Others create support load, churn risk, or an access-control problem that only becomes visible during an audit.
For an operator, the useful definition is operational: revenue leakage is the set of cases where billing state and entitlement state disagree in a way that changes the customer experience or the economics of the account. The disagreement can favour the customer, favour the business, or be genuinely ambiguous.
Four categories worth separating
- Paid, but blocked: an active or trialing subscription exists, but the application does not grant the expected access. This is a retention and support risk. It can also be a refund risk.
- Access without a valid subscription: the application still grants premium access after a cancellation, failed payment, or expiry. This is the classic revenue leak, but the cause may be a legitimate manual grant, a grace policy, or a delayed webhook.
- Plan or seat mismatch: the customer pays for one tier and receives another, or pays for three seats while ten users have access. The billing record and the entitlement record are both present, but they describe different products.
- Unmapped records: a subscription or entitlement exists without a reliable identity link. The gap may be benign (a deleted test customer) or material (a paying account nobody can find).
Why a single “revenue at risk” number is dangerous
It is tempting to multiply every stale entitlement by a monthly price and present the result as recovered revenue. That is usually dishonest. A canceled subscription may still be inside a paid-through period. A manual grant may be intentional. A missing Stripe record may be caused by a partial export rather than a missing payment.
A defensible estimate needs three ingredients: a verified price for the relevant subscription, a known access state during a defined time window, and a documented policy for whether the case should count. If any ingredient is missing, label the amount unknown. That is more useful than a precise number built on assumptions.
The minimum evidence trail
Every finding should point back to the records that produced it: the Stripe customer and subscription identifiers, the entitlement row and user identifier, the snapshot timestamps, and the policy version used for the comparison. This is what lets a second person review the finding without rerunning the entire audit.
The audit should also record whether each source snapshot was complete. Reconciling against a partial Stripe export is worse than skipping the run: it can turn every missing row into a false “access without payment” finding.
SubSync Guard is deliberately advisory. It produces reviewable signals; it does not revoke access, cancel subscriptions, or issue refunds.
A practical starting point
Start with a read-only export of subscriptions and a read-only export of entitlements. Normalize customer identity, exclude obviously incomplete snapshots, and compare states under a written policy. Review the first twenty findings by hand before you automate anything. The patterns you see will tell you whether the real problem is billing, provisioning, identity mapping, or all three.
Primary references