What is revenue leakage? A stage-by-stage breakdown

Losing a customer to a competitor is a strategy problem. Losing a customer because their card expired on a Tuesday and your retry fired at 3am in their timezone is an operations problem.

Almost every company has a plan for the first one. Very few have a view of the second.

Revenue leakage is money a business has already earned the right to collect and loses to preventable causes — failed payments, silent product regressions, discount habits, journey breaks, support failures — rather than to competitive loss or genuine customer choice.

This post is the map: ten stages, what leaks at each, who owns it, and how you'd tell.

Key takeaways

• Leakage is preventable loss, not competitive loss. Separating the two changes what you do on Monday morning.
• The leak rarely appears in the stage where it originates. Churn shows up in Renew; the cause usually lives in Activate, Support, or Price.
• Most leakage is invisible in a dashboard because dashboards aggregate. A four-point drop in one market inside a flat global number is a leak nobody sees.
• Every leak has an owning team, and the long-running ones have three. Leaks live in the seams between teams.
• A leak with no dollar figure will not get prioritized. Denominate everything, and split it by segment before you report it.

Why leakage hides

Aggregation. A global repeat-purchase rate moving from 34% to 33% looks like rounding. Split by segment it can be flat in enterprise, flat in mid-market, and down eleven points in self-serve. The aggregate is a hiding place.

Misattribution to the wrong stage. When a customer doesn't renew, the number lands in Renew and Customer Success gets asked about it. But the reason might be an activation failure eight months earlier. Where a loss is recorded is rarely where it was caused.

No owner for the seam. Every company has an owner for checkout and an owner for delivery. Almost none has an owner for what happens when a checkout change alters how the delivery fee is displayed.

Strong association across the businesses we work with, the leaks that run longest are consistently the ones crossing two or more team boundaries. Single-team leaks get caught fast, because someone's number moves and they own it.

The ten stages

1. Acquire — Marketing

Leaks as: channels that look efficient at day 30 and are unprofitable at month 6; CAC measured against first order rather than contribution; cohorts acquired on discount that never repeat at full price.

How you'd tell: cohort acquisition by channel, then look at M6 retention and contribution rather than signups. A channel with excellent CAC and poor M6 retention is a leak wearing a growth costume.

Also implicates: Finance, because the discount that made the channel look good was a pricing decision.

2. Activate — Product

Leaks as: customers who pay and never reach first value. Every one is a refund, a churn, or a support ticket on a delay.

How you'd tell: define first value concretely — first order delivered, first workflow completed, first teammate invited — then measure what share of new customers reach it and where the rest stop.

Also implicates: Support, because the drop-off point appears in tickets weeks before it appears in a metric.

3. Price — Finance

Leaks as: discount dependency. A discount given once to close becomes a discount expected at every renewal. Margin erodes because no single approval looks unreasonable.

How you'd tell: track the share of revenue transacted at full price over time, and cohort customers by whether their first purchase was discounted. The lifetime margin gap is usually stark.

Also implicates: Sales and Marketing, who both have local incentives to discount.

4. Adopt — Product

Leaks as: customers paying for capability they never touch. In B2B that's a renewal risk with a countdown on it. In B2C it's subscriptions that lapse without complaint.

How you'd tell: compare entitlement against usage. A customer on your top tier using only bottom-tier features is a downgrade already scheduled.

Also implicates: Customer Success, who need this months before renewal.

5. Retain — Marketing

Leaks as: repeat-purchase decay. In most consumer businesses the second-order rate is the single most sensitive early indicator, and it responds to things marketing doesn't control — delivery times, fee presentation, stock availability, packaging.

How you'd tell: cohort by first-order date, plot second-order conversion. A step change aligned to a dated deploy is a causal finding waiting to be confirmed.

Also implicates: Product and Engineering, because the cause is frequently a release.

6. Expand — Sales

Leaks as: upsell signals that expire unseen. An account hitting usage limits is telling you it's ready; unseen, it either downgrades in frustration or plateaus.

How you'd tell: instrument the thresholds. Approaching a plan ceiling should generate a signal, not a surprise invoice.

Also implicates: Product, who own the limit, and Finance, who own the overage policy.

7. Support — Support

Leaks as: systemic issues handled as individual tickets. Support is measured on resolution time, which optimizes for closing tickets quickly rather than noticing four hundred of them share a cause.

How you'd tell: cluster tickets by reason, then cross-reference against repeat-purchase behaviour. Customers who contact you about a specific issue and never return are quoting you the price of that issue.

Also implicates: Product and Engineering, who own most systemic causes.

8. Renew — Customer Success

Leaks as: involuntary churn. Cards expire, banks decline, retries fire at the wrong hour. These customers did not choose to leave, they are the cheapest revenue in the business to recover, and they are usually counted inside the same churn number as people who left deliberately.

How you'd tell: split churn into voluntary and involuntary before doing anything else. If you can't split it, that's the first thing to fix.

Also implicates: Finance and Engineering, who own the payment stack and the retry logic.

9. Advocate — Marketing

Leaks as: referral behaviour that happens but isn't instrumented, so it can't be encouraged — and so paid CAC efficiency is overstated.

How you'd tell: attribute acquisition back to referring customers where the data supports it, and mark it unavailable where it doesn't.

10. Churn — everyone

Leaks as: the framing itself. Churn is a symptom recorded at the end of a lifecycle. Treating it as a stage with an owner produces win-back campaigns aimed at people whose reason for leaving was set months earlier.

How you'd tell: attribute every churn back to its source stage. A churn caused by a failed payment belongs to Renew. A churn caused by never reaching first value belongs to Activate. Preventable versus natural is the split that matters.

What this looks like in practice

The pattern we see most often in consumer businesses runs like this.

Second-order rate drops. It surfaces in Retain, so Marketing gets asked. Marketing runs a win-back, which converts poorly — because the customers aren't hesitant, they're annoyed. The actual cause is a checkout change that surprised them with a fee at the last step.

Causal finding needs three things here: the dated deploy, the cohort exposed to it, and a comparison cohort that wasn't. With all three you have a cause. Without the dated deploy you have a strong association, and you should say so rather than dress it up.

The fix isn't a campaign. It's a product change — owned by Product, surfaced by Support, quantified by Finance, coordinated by whoever owns the number. Four teams. Which is precisely why that leak survived nineteen weeks.

What we can't tell you yet

There's no reliable industry benchmark for total leakage as a percentage of revenue, and I'd be sceptical of anyone quoting one confidently. It varies enormously by business model, payment mix, and market.

We also can't tell you which stage is worst in your business without your data. Anyone who can is guessing.

Frequently asked questions

What's the difference between revenue leakage and churn? Churn is one outcome of leakage. Leakage also covers margin lost to discounting, revenue lost to failed payments that never became churn, refunds, and expansion that never happened. Churn is the visible tip.

How is involuntary churn different from voluntary churn? Involuntary churn is a payment failure — expired card, bank decline, badly timed retry. The customer never decided to leave. Voluntary churn is a decision. They need completely different responses, and most churn dashboards combine them.

Which stage should I instrument first? Renew. Involuntary churn is usually the fastest money to recover, and splitting voluntary from involuntary is one-time work. Then Retain, then Price.

Why do you refuse to show one blended total? Because a single number hides which segment or market is actually leaking. We break every figure out by segment and market, and for cross-border customers we hold each market in its own currency rather than converting.