Most subscription operators read their P&L top down: MRR, churn, gross margin, CAC payback. That framing hides the single most expensive line in the business, which is the gap between revenue you are owed and revenue you actually collect. In a recurring model, billing is not an accounting formality at the end of the month. It is an operational system that fails a measurable percentage of the time, and every failure is margin you already earned walking out the door.
This module sets up the rest of the course. Before we touch dunning, card lifecycle, or disputes, we need a shared mental model of the recurring-revenue P&L and exactly where it leaks. Get the map right and every later module becomes a targeted fix rather than a general best practice.
Billed revenue is not collected revenue
In a one-time commerce model, a failed payment means a lost cart. The customer either retries or leaves, and you never recognized the revenue in the first place.
In a subscription model the relationship is inverted. You have a paying customer, an active entitlement, and a renewal that is supposed to fire automatically. When the charge fails, you have already delivered the product for the prior period and you are about to keep delivering it. The revenue was real. The collection failed.
This is why the headline churn number lies. Reported churn blends two very different things: customers who chose to leave (voluntary) and customers whose payment broke (involuntary). The second group never decided anything. They are leaking out of a hole in the plumbing.
Across the subscription industry, involuntary churn typically accounts for roughly 20 to 40 percent of total churn, and an average subscription business loses on the order of 9 percent of monthly recurring revenue to failed payments before any recovery effort (Butter Payments, FlyCode). For a business at $10 million ARR, that is a seven-figure line item that never shows up as a line item.
The four leak points
We find it useful to break the recurring-revenue P&L into four distinct leaks. Each has a different owner, a different fix, and a different module later in this course.
1. The decline at renewal
The first leak is the charge that the issuer refuses. Network data and processor benchmarks put recurring-payment decline rates in the rough range of 10 to 15 percent on initial attempt, climbing higher for subscription-heavy merchants (Chargebacks911).
The split matters more than the headline. Most declines are soft declines, temporary refusals such as insufficient funds or a velocity hold, often cited at 80 to 90 percent of all declines on valid cards. The remainder are hard declines: expired, closed, or flagged accounts. Soft declines are recoverable with retry timing and logic. Hard declines need a new credential, which is a different fix entirely. The dunning engine (module 3) handles the soft side; card lifecycle management (module 4) handles the hard side.
2. The stale credential
The second leak is the card that no longer exists. The average payment card is valid for three to four years, and expirations plus reissues mean a meaningful slice of your stored credentials are quietly going bad every month. Expired cards are commonly cited as one of the largest single drivers of subscription payment failure.
Network account-updater services (Visa Account Updater, Mastercard Automatic Billing Updater) refresh stored credentials, but they do not catch everything. A realistic hit rate on eligible expired-card failures is roughly 60 to 80 percent, not 100 percent, and prepaid and some regional cards may not be enrolled at all (Gr4vy). Treating the updater as a complete fix is a common and expensive mistake.
3. The voluntary cancel you could have saved
The third leak is genuine churn, but a portion of it is recoverable through offboarding flow, pause options, and downgrade paths rather than a hard cancel. This is the leak closest to product and pricing, and it is the one operators tend to over-index on because it feels like the "real" churn. It is real, but it is usually smaller than the involuntary leak.
4. The chargeback on a legitimate charge
The fourth leak is the dispute. A customer who forgot they subscribed, or who could not find the cancel button, files a chargeback instead. You lose the transaction, you pay a dispute fee, and the dispute counts against your acquirer monitoring thresholds. This is friendly fraud against a recurring charge, and it is the subject of modules 6 through 9. The general four-party dispute system is covered in the sibling course; here we stay anchored to the recurring case.
A worked example
Take a DTC subscription doing $1,000,000 in billed MRR, AOV around $40, mostly monthly card-on-file renewals.
- Renewal declines at 12 percent put $120,000 of billed MRR at risk each month before any recovery.
- Of that, assume 85 percent are soft declines ($102,000). A competent retry and dunning program recovers perhaps half, so roughly $51,000 comes back and $51,000 leaks.
- The remaining 15 percent, about $18,000, are hard declines needing a new credential. An account updater at a 70 percent eligible hit rate saves around $12,600 and leaks roughly $5,400.
- Voluntary cancels and recurring chargebacks run separately on top.
Net involuntary leak in this single month sits near $56,000, or about 5.6 percent of billed MRR, recoverable to a large degree but only with the systems in modules 3 and 4 actually built and tuned. Annualize it and the gap dwarfs most pricing experiments the team is debating.
Why this is the highest-leverage view
Net revenue retention is the metric investors anchor on, with a median net revenue retention in the high 100s, around 106 to 108 percent, and gross revenue retention in the high 80s for B2B SaaS (SaaS Capital). The fastest path to lifting both is rarely a new expansion motion. It is plugging the involuntary leak, because that revenue is already sold, already delivered, and already counted as lost in your churn number.
The takeaway for this module: separate billed from collected, split your churn into voluntary and involuntary, and assign every dollar of the involuntary slice to one of the four leak points. The rest of the course is the toolkit for each. If you cannot yet produce a clean number for involuntary churn this month, that is the first thing to fix, because you cannot recover a leak you are not measuring.