The two leaks

Why fixing your billing won't fix your forecast.

Search the phrase "revenue leakage," and you will get pages of billing software. Quote-to-cash platforms, revenue-assurance tools, invoice reconciliation, contract compliance — a whole mature category, with analyst coverage, CFO guides, and case studies of recovered money. All of it real, all of it useful. And almost none of it is about the leak this site is concerned with.

That's because B2B uses one phrase for two different leaks — and the one with the software, the statistics, and the CFO's attention is the smaller, bounded one. The other leak is larger, sits upstream of everything the tooling can see, and has no product category at all. Confusing them has a cost: companies buy the fix for the first leak, report the topic handled, and keep paying for the second in full.

This essay is the sorting exercise: what each leak is, what each is worth, and how to tell which one is showing up in your numbers.

Leak one: earned but not collected

The first leak lives between contract signature and cash. The mechanisms are well documented: usage that was delivered but never billed, pricing errors, and legacy rates that survive renewals, contract escalators nobody enforces, invoices delayed or missed when data falls between the CRM and the billing system, and discounts applied beyond what was approved. Every one of these is money the company earned — the work was done, the service delivered — that never arrived.

I assumed, going in, that this leak at least had honest numbers — it's the measured, tooled, finance-owned one, after all. Then, I did what the first essay demands and walked the citations back. The most quoted figure — companies lose between 1 and 5 percent to this kind of leakage, attributed to EY — traces through a Forbes article to a dead link on EY's own site, and the denominator mutates along the way: EBITA in one retelling, EBITDA in most others, plain revenue in the rest. Its companion statistic — 42 percent of companies experience some form of leakage, attributed to MGI Research — resolves to MGI's general research page and no locatable study. MGI is a real, independent research and advisory firm serving business, finance, and technology executives; the specific number, wherever it began, now circulates without its evidence. So, the finding stands, uncomfortably, on both sides of the fence: even the well-tooled leak runs on laundered statistics.

Here is what's genuinely different about Leak One, and it was never the quality of its benchmarks: this leak is reconcilable. The contract says one number, the invoice says another, the bank account says a third, and the leak is the delta between your own records. You don't need to believe anyone's percentage — you can measure it directly, this quarter, from your own ledger. It is also the bounded leak: it can never exceed the revenue you contracted and failed to collect. Reconcilability is precisely why software can find it — reconciliation is what software does best — and why the recovered money shows up fast and lands almost entirely on the bottom line, since the cost of serving those customers was already spent. If you run complex contracts on manual billing, this category deserves your money. I don't do that work, and I'm glad someone does it well.

Leak two: never earned at all

Now read the tooling category's own definition, because it draws the boundary for us. Revenue leakage, the glossaries say, is money a business earned but didn't collect — as distinct from revenue loss, income never earned in the first place. Everything upstream of the signed contract is, by the category's own definition, out of scope.

That out-of-scope territory is where the second leak lives: in the handoffs between marketing, sales, and customer success. The lead that cooled in the queue between two definitions of "qualified." The deal that stalled between demo and proposal. The expansion that never happened because onboarding delivered less than the proposal promised. A long-standing IDC estimate puts the cost of this misalignment at 10 percent or more of revenue per year — an estimate, as I've argued before, not a measurement of you, but a credible prior for how big the territory runs.

And the structural property flips. Leak two is not reconcilable, because there is nothing to reconcile: a deal that was never created produces no record, and a lead that died of queue time looks identical in the CRM to a lead that was never any good. No scanner exists for what didn't happen. That is why there is a thriving software category for the bounded leak and no product category for the big one — software audit records, and the big leak's defining feature is the absence of them. It is also why the two leaks respond so differently to bad statistics. Leak one can shrug off its laundered benchmarks: your own reconciliation replaces them with a real number in a quarter. Leak two has no self-serve number to fall back on — which is exactly the vacuum the laundered percentages rush to fill, and why the only honest alternatives are a labeled prior or a purpose-built measurement.

Why is the confusion expensive

The cost of using one phrase for both is a quiet box-ticking error. A board raises "revenue leakage"; finance procures a billing or revenue-assurance platform; collections improve, EBITDA firms up — genuinely — and the topic is filed as handled. But look at what the fix could never touch. Billing hygiene recovers revenue from deals you already won. It cannot move the forecast, because forecast misses come from the pipeline that doesn't behave as recorded — an upstream problem. It cannot bend the cost of growth, because that is spent leaking between functions before any contract exists. It cannot lift expansion, because expansion dies in the handoff to customer success, years before any invoice goes wrong. Fixing your billing won't fix your forecast — not because the billing fix failed, but because it was aimed at the other leak.

So the triage is worth two minutes. If your symptoms are deltas — contracted-versus-billed gaps, invoice disputes, missed escalators, creeping days-sales-outstanding — you have leak one; buy the tooling. If your symptoms are the other set — the forecast misses while every function's dashboard is green, growth costs more each euro every year, renewals arrive with quiet discounts, the same big deals roll forward quarter after quarter — you have leak two, and no amount of billing automation will reach it. Most companies past a certain complexity have some of both. They are different diseases, and they take different instruments.

Two leaks, two instruments

The first instrument exists and is mature: reconciliation software, plus a finance owner. The second instrument cannot be software-first, for the reason above — it has to construct the picture that no system recorded, by reading across the records that do exist: what the CRM shows, what the financials show, what the people at each handoff know, scored on one scale that stays constant so the answer can be re-measured after the work. That is the instrument I built, and the honest division of labor is worth stating plainly: the billing category protects revenue you've earned; engine measurement finds the revenue you're set up never to earn.

Both leaks are real. Only one of them has a queue of vendors offering to fix it. So, the next time "revenue leakage" comes up in a leadership meeting, ask one clarifying question before anyone opens a budget: which leak do we mean — the money we earned and lost, or the money we never earned at all? The first has a software category waiting. The second has been waiting for you.

Want the second leak measured for your own engine? Start with a €750 COS Efficiency Diagnostic.