Most finance teams don’t “lose money” in one dramatic event. They lose it in small, repeatable cracks — the gap between what was agreed, what was delivered, what was billed, and what was collected.
That’s why “leakage” is so stubborn: it hides inside day-to-day operations, not inside a single KPI.
A useful reference point: EY has estimated companies can lose up to ~5% of earnings through revenue leakage. (Stripe) And in real life, a big chunk of payment delays are not “bad customers” — they’re process friction: UK government research found businesses often attributed late payments to administrative errors (36%) and disputed invoices (31%). (GOV.UK)
Why this matters in 2026
Even if your top line is growing, leakage quietly:
- delays cash (DSO drift)
- increases bad debt exposure
- forces finance into rework loops (disputes, credit notes, reconciliations)
- creates control/audit headaches
For example, Atradius’ Western Europe barometer reported 47% of B2B invoices overdue, and noted bad debts affecting ~6% of B2B invoices on average (a meaningful hit when you scale). (Atradius)
9 leakage zones from contract → invoice → cash
1) Contract terms never become executable
Discounts, minimum commitments, indexation, renewal uplifts — agreed in PDFs, not translated into structured billing logic. This is “silent leakage” because everyone assumes the system is doing it.
2) Invoices go out before the work is “clean”
Missing PO, wrong entity, wrong tax treatment, wrong ship-to / bill-to, wrong customer reference → disputes → delayed cash. And disputes aren’t rare; they’re a known driver of late payments. (GOV.UK)
3) Credits and claims don’t close the loop
A credit is “approved” in email, but not issued. Or issued, but not matched to the original dispute. That turns into ageing A/R and write-offs.
4) Usage / delivery and billing logic drift apart
If delivery data, pricing logic, and invoicing don’t reconcile cleanly, the business runs on “best effort billing.”
5) Renewals and uplifts are run on calendar reminders
The renewal happens. The uplift doesn’t. This one is brutal because it’s pure margin.
6) Write-offs are treated as “collections outcomes,” not process signals
Write-offs are often the final symptom of upstream errors (incorrect terms, wrong documentation, repeated disputes). McKinsey notes that tightening write-off thresholds and identifying non-genuine write-offs can recapture 40–60% of value leakage in some organisations. (McKinsey & Company)
7) Multi-entity and FX introduce quiet variance
Wrong entity routing, inconsistent rates, rounding, intercompany mismatches — small errors that accumulate.
8) Manual reconciliations hide root causes
If finance is “reconciling reality” every month, the process is broken upstream — you’re paying people to be a human middleware layer.
9) Ownership gaps across the chain
Everyone owns a step. Nobody owns the outcome (contract → invoice → cash). This is where leakage becomes cultural: “not my queue.”
What good looks like (a practical framework)
1) Make terms executable, not readable
The goal isn’t better document storage. It’s turning commercial terms into structured data that downstream processes can enforce.
2) Put control points where errors scale
You don’t need more approvals — you need the right approvals:
- before billing goes out
- when exceptions occur
- when credits are issued
- when write-offs are requested
3) Run an exception-first operating model
Routine should flow; exceptions should route to humans with full context.
4) Close the loop (prevention > clean-up)
Every dispute/credit should create a prevention rule or a process fix — not a spreadsheet note.
This is why “measurement” matters: APQC explicitly frames O2C measurement as foundational to improvement (you can’t fix what you don’t instrument). (APQC)
The metrics that actually expose leakage
Beyond DSO (definition + calculation are standard), track the metrics that reveal why DSO drifts. (Investopedia)
- Invoice dispute rate (and top 3 root causes)
- Credit-note cycle time (request → approval → issued → matched)
- First-time-right invoicing (error-free invoices the first time) (APQC)
- Write-offs by root cause (not just by amount)
- Duplicate/erroneous payment rate (AP side leakage exists too; APQC benchmarks are widely used here) (APQC)
Start this week (3 moves that create real momentum)
- Pick one leakage stream and measure weekly Choose one: disputes, credits, renewals/uplifts, write-offs, FX/entity errors. Make it visible.
- Define 3 control points that prevent 80% of errors Where are errors born? Put guardrails there (not at the end of the month).
- Track exceptions per invoice Not volume. Not “hours saved.” Exceptions are where cost and cash hide.
At Farben, we’re building an agent layer that helps finance teams detect leakage and drive controlled execution (human approvals + audit trails), so outcomes show up in cash — not just in dashboards.
If you find our articles useful, consider subscribing!