Why this number is missing from most board decks
Churn dashboards almost always report a single blended figure. That figure blends two completely different phenomena: customers who chose to leave, and customers whose payment mechanics broke. The first is a product and pricing problem measured in quarters. The second is an operations problem measured in days.
Blending them hides the cheapest retention win available to a subscription business. Recovering involuntary churn requires no product change, no discount, no price experiment and no additional acquisition spend — the customer already decided to pay.
The first job, then, is a split. Tag every churn event by cause: cancellation requested, or subscription terminated after payment failure. If your billing platform does not expose this cleanly, derive it from whether the subscription passed through a past-due or unpaid state before ending.
Step one: your true failure rate
Count failed renewal attempts as a share of total renewal attempts over a full billing cycle, not a calendar month, because renewal dates cluster and a calendar cut will misstate both sides.
Typical ranges by model: B2B SaaS 4-7%, digital publishers and memberships 6-12%, consumer ecommerce subscriptions 8-14%. Business cards fail less often than consumer cards, and annual plans fail less often than monthly ones simply because they attempt less.
If your measured rate sits far outside these bands, check your denominator before you celebrate or panic. Excluding retries from the count is the most common error and it can halve the apparent rate.
Step two: at-risk MRR, weighted properly
At-risk MRR is not failure rate multiplied by total MRR. Failures are not uniformly distributed across price points, and in most portfolios the failure rate on the cheapest tier is meaningfully higher than on the most expensive one.
Compute it directly: sum the amounts of all failed renewal attempts in the cycle, deduplicated per subscription so three attempts on one $99 plan count once. That sum, expressed monthly, is your at-risk MRR.
Annualise carefully. At-risk ARR is not at-risk MRR times twelve, because a recovered subscription continues while a lost one stops — the correct figure is the lifetime value forgone, which is always larger than the single missed invoice.
Step three: realistic recovery bands
Recovery varies enormously by decline code, so a single assumption produces a meaningless forecast. Use the observed bands: expired card 70-85%, insufficient funds 55-70%, authentication required 60-80%, do-not-honor 30-45%, and terminal codes such as lost or stolen cards 20-40%.
Weight those bands by your actual code mix. A B2B portfolio heavy in expired corporate cards will model far better than a consumer portfolio dominated by short balances, and both will disappoint if you assume a flat 60%.
Then subtract your baseline. Your payment platform's own retry logic already recovers a share of failures with no help. Only the increment above that baseline is attributable to a recovery programme, and any vendor forecast that skips this step is overstating the case.
Step four: cost per recovered dollar
This is the metric that separates pricing models. Under a flat monthly fee, cost per recovered dollar falls as recovery grows: a $250 plan recovering $18,000 a month costs about 1.4 cents per dollar. Under a nine percent revenue share, the same recovery costs $1,620 — roughly nine cents per dollar, permanently, at every volume.
Add the variable channel cost. WhatsApp conversations are billed per conversation by Meta directly and email is fractions of a cent, so on realistic volumes the messaging line stays small relative to either pricing model.
Finally add engineering time if you are building rather than buying: webhook verification, idempotency, scheduling, template management, deliverability and reporting is a multi-week build and a permanent maintenance obligation.
Step five: payback and sensitivity
Payback is usually measured in days rather than months. If a $250 monthly plan recovers even one $300 annual subscription in the first week, the tooling has paid for itself for the month.
Run the model at three recovery levels — pessimistic, expected, and strong — using the low, mid and high ends of the weighted bands. If the pessimistic case still clears cost by a wide margin, the decision is not close and does not need more analysis.
The sensitivity that matters most is not recovery rate; it is code mix. Re-run the model with a five point shift toward do-not-honor and see how much of the case survives. If it collapses, your programme design needs an alternate-card strategy before it needs a budget.
How much of your involuntary churn is recoverable?
Compares a 40% single-channel baseline against the 63.8% RRLabs platform average.
- At risk / month
- $5,600
- Extra recovered / month
- $1,333
- Annualised, less $3,000 plan
- $12,994
RRLabs vs legacy dunning tools
Churn Buster, Baremetrics Recover, Stripe native dunning and Gravy compared with Revenue Recovery Labs.
| Feature | Legacy tools | Revenue Recovery Labs |
|---|---|---|
| Recovery channels | Email and basic SMS only | Native Meta WhatsApp Cloud API (your number) + email |
| Pricing architecture | $129–$629+/mo, or a percentage cut of recovered cash | Flat $100 / $250 / $500 per month — you keep 100% |
| Retry logic | Fixed calendar dunning (day 1, 3, 7) | Retry window derived per decline code, persisted at ingestion |
| Copy engine | Static templates you edit by hand | 4-tier cascade: cache → primary model → fallback model → deterministic templates |
| White-label | Unavailable or enterprise-only | Flat +$300/mo with custom domain and no vendor watermark |
| Audit trail | Send logs at best | Tier, model, latency and recovery score on every message |
What to put in front of the board
Four numbers on one line: at-risk MRR, baseline recovery from the platform alone, incremental recovery from the programme, and cost per recovered dollar. Everything else is supporting detail.
Add one chart: involuntary versus voluntary churn over the last twelve months, stacked. The moment those two are visually separated, the conversation about which one to fix first becomes obvious rather than contested.
Commit to reporting the same four numbers monthly. Recovery programmes decay quietly — a broken update link, an expired template, a deliverability slip — and a standing metric is the only reliable early warning.
Common modelling mistakes
Counting gross recovered revenue without subtracting the platform's own baseline recovery. This is the single largest source of overstated cases.
Using a flat recovery assumption across all decline codes, which produces forecasts that are simultaneously too optimistic for consumer portfolios and too pessimistic for B2B ones.
Ignoring lifetime value. A recovered subscription is not worth one invoice; it is worth the remaining expected life of the relationship, which is why involuntary churn recovery usually outperforms every acquisition channel on a per-dollar basis.
Forgetting the cost asymmetry of over-messaging. Aggressive dunning converts involuntary churn into voluntary churn, and that cost never appears in a recovery dashboard.
A worked example
Take a $60,000 MRR B2B SaaS business with a 6% first-attempt renewal failure rate. At-risk MRR is roughly $3,600 per cycle. Its code mix is 45% expired card, 25% insufficient funds, 15% authentication required and 15% do-not-honor.
Weighting the mid-points of the observed bands gives an achievable recovery of roughly 62%, or about $2,230 per month — before subtracting the baseline the payment platform's own retries already deliver, which might be a third of that.
The incremental figure, call it $1,500 per month, is what a recovery programme is actually worth here. Against a $250 flat monthly plan that is a six-fold return in the first month, and the comparison improves every time MRR grows because the fee does not.
Lifetime value, not just the missed invoice
The worked example above understates the case, because a recovered subscription keeps paying. If average remaining customer life is eighteen months, each recovered $200 subscription is worth $3,600 of future revenue, not $200.
Model both figures and present both. The invoice-level number is the conservative one finance will trust; the lifetime number is the one that reflects reality.
The same asymmetry applies to over-aggressive dunning. A cancellation triggered by pressure costs the full remaining lifetime value, which is why complaint rate belongs in the model alongside recovery rate.
Where RRLabs fits
Revenue Recovery Labs reports at-risk revenue, recovered revenue, recovery rate by decline code and time-to-recovery on one dashboard, so the numbers in this model come out of the system rather than out of a spreadsheet.
Pricing is flat at $100, $250 or $500 per month with a $300 white-label add-on, and no percentage is ever taken from recovered revenue — which is the single largest variable in the cost-per-recovered-dollar calculation above.
The interactive calculator on this page uses the same weighted-band method described here, so you can size your own recoverable MRR before committing to anything.