Explainer

October 12, 2026·13 min read

Written by Darya Birukova

Revenue Leakage in Logistics: The Losses Your Billing Audit Won’t Find

Revenue leakage in logistics usually gets treated as a billing problem: accessorials nobody invoiced, surcharges applied at the wrong rate, contract terms that drifted. Those leaks are real, and a good freight audit finds many of them. But there is a second kind an audit can’t see, because it never produces a billing error: the order that was cancelled, the delivery that failed, the demurrage that piled up before anyone noticed.

This guide covers both billing and operational kinds of revenue leakage. It explains how big the problem is, why the operational kind hides from standard reports, where it typically sits in forwarding, road haulage, warehousing and last-mile operations, and how to find it in your own data.

Key takeaways

• Billing audits only verify charges against executed work and cannot detect revenue lost from work that never happened.
• Operational leaks stem from root causes split between disconnected systems, such as order management, chat logs, and driver rosters.
• Aggregated monthly reports disguise significant operational spikes that only become visible at lane, weekly, or customer levels.
• Systematic operational leaks with controllable causes and clear operational ownership should be prioritised over visible one-off issues.

What is revenue leakage in logistics?

Revenue leakage is money a logistics business should have kept but did not, falling into two distinct categories: billing leakage and operational leakage.

Billing leakage represents work that was performed but never fully invoiced, or invoiced but never collected. Examples include a driver waiting three hours without a detention charge reaching the invoice, a re-labelled pallet missing a handling fee, or a fuel surcharge calculated at an outdated quarterly rate. Finance departments and freight audits uncover these gaps because a clear discrepancy exists between logged activity and the generated invoice.

Operational leakage is revenue lost, or margin consumed, by how the operation actually runs. This includes orders cancelled because no driver was available, containers sitting past allowable free time due to a customs document trapped in an email inbox, trucks running empty return legs without a matched backhaul, or carrier invoices paid without validation against actual delivery records. These losses show up nowhere obvious on financial reports, often appearing as normal overtime, standard cancellations, or correctly calculated third-party charges.

While billing leakage is addressed by invoice audits, operational leakage requires root-cause analysis across operational systems.

How big is revenue leakage in logistics?

There is no reliable all-industry percentage, and we’d be wary of anyone quoting one. Leakage depends too much on the segment, contract structure, and how an operation’s data is split. What does exist is solid research on individual leaks, and the individual numbers are large.

Take truck detention in the US. In its 2024 study, the American Transportation Research Institute (ATRI) found drivers were detained in 39.3% of stops in 2023. Almost every fleet (94.5%) charges detention fees, yet fewer than half of those invoices get paid. ATRI put the cost to the for-hire industry at $3.6 billion in direct expenses plus $11.5 billion in lost productivity for that one year.

That single finding contains both kinds of leakage. The unpaid detention invoices are billing leakage: the charge existed, the money never arrived. The 135 million hours of lost driver time are operational leakage: capacity that could have moved freight and didn’t.

In Europe, the clearest structural example is empty running. Eurostat’s road freight statistics show that roughly a fifth of all road freight vehicle-kilometres in the EU are driven by empty vehicles, with national transport running empty far more often than international. Not every empty kilometre is avoidable. But each one is paid for in fuel, wages and wear, and earns nothing.

Neither figure tells you what your own operation is losing. They tell you that leaks of this size are normal, mostly unmeasured, and rarely visible on a single report.

Why doesn’t a billing audit catch operational leaks?

A billing audit compares what happened against what was invoiced. It’s good at finding activity without a matching charge. It has nothing to compare against when the revenue-generating activity never happened.

There are three reasons operational leaks slip through:

1. There is no event to compare. A cancelled order leaves no delivery to bill, so there’s no gap for the audit to flag. A demurrage invoice from a terminal is correctly calculated and gets paid. The loss is real, but every document involved is accurate.
2. Standard reporting averages it away. Most operational dashboards report by month and by average. A cancellation rate that jumps for three weeks on two lanes, then settles, barely moves the monthly figure. Weekly spikes that would be obvious at lane level disappear in the total.
3. Symptom and cause live in different systems. Cancellations sit in the order system. The reason for them sits in driver schedules, a carrier’s WhatsApp thread or a planner’s inbox. Unless someone joins the two, the cause stays invisible.

Where does revenue leak in a logistics operation?

Operational and billing leaks occur at several distinct touchpoints across execution, dispatch, and settlement:

Accessorial charges earned but never billed

Waiting time, extra stops, re-deliveries, tail-lift use, out-of-hours handling: these accessorials are usually earned on the road and recorded, if at all, in a phone call or a chat message. If the evidence never reaches the billing system, the charge never goes out. If it goes out without evidence, it’s disputed and often written off, which is part of what the ATRI payment gap reflects.

There’s a telling detail in the same ATRI research: GPS data showed trucks drove faster on trips to facilities where they were later detained. Drivers know which sites will hold them up. That knowledge sits in people’s heads, not in a system that could price it, plan around it or bill for it.

Unexamined cancellations

Every operation has a cancellation rate, and most have reason codes. Look at how many cancellations end up coded “other”, “customer request” or left blank. That is usually where a systematic cause gets filed as noise. When cancellations rise, the useful question isn’t how many, but why these orders, on these lanes, on these days.

Premature demurrage and detention payments

Free time runs out quietly. A container waits at the terminal because the customs document is still in someone’s email. A truck sits at the dock because a slot was confirmed by phone and never entered. By the time the charge arrives, it’s a correct invoice for a problem that was preventable days earlier. Forwarders and drayage operators often treat these fees as a cost of doing business; a share of them is a recurring process failure with a traceable cause.

Unproductive empty kilometres

Empty running is partly structural: some lanes are one-directional and some return legs will never fill. But part of it is coordination. A return load was available and nobody matched it in time, or a re-delivery sent a truck back over the same route. These kilometres rarely appear as a loss; they appear as normal fuel and driver cost.

Untracked operational re-work

Failed deliveries, re-deliveries, manual corrections and repeat calls to fix the same order rarely get their own line in the P&L. They show up as overtime, extra kilometres and a dispatcher’s afternoon. Each instance is small. A recurring pattern on the same customer or lane is margin leaking every week.

Unaudited carrier invoices

This sits between billing and operations. When a carrier invoice isn’t reconciled against the order and proof of delivery before payment, overcharges go through. We model up to ~5% of invoice overpayments as preventable with automatic three-way reconciliation; the real figure for any operation only comes out of its own data.

Delayed cash collection

Not every leak is a loss; some are delays. If invoices go out three to five days after delivery because someone has to assemble the paperwork by hand, cash arrives later on every order. That doesn’t show as lost revenue, but it is working capital the business funds itself. We model DSO reductions of up to ~10% when invoices are issued on the day of delivery.

Which leaks matter most for your type of operation?

The same leak types appear everywhere, but their weight differs by segment. As a rough guide from what we see:

• Ocean, air and multimodal forwarders: demurrage and detention exposure, document delays that hold cargo, and accessorials lost between partners.
• Road haulage and trucking carriers: unbilled waiting time, empty kilometres, and cancellations driven by capacity gaps.
• Drayage and port logistics: free-time overruns and gate congestion, where a few hours’ delay converts directly into fees.
• Warehousing and distribution centres: no-show slots, unbilled handling and value-added services, and detention charged back by carriers.
• E-commerce, last-mile and cold-chain 3PLs: failed deliveries, cancellations and re-delivery cost, where volume makes small rates add up fast.

Use this as a starting hypothesis, not a conclusion. The biggest leak in a given operation is often not the one its management expects.

What does an operational leak look like in real data?

In our own work with a leading US last-mile delivery operator, we analysed around 1.3 million orders. Their standard reports showed nothing unusual. The analysis surfaced a roughly 50% spike in order cancellations and traced it to its root cause: a driver shortage.

None of the inputs were hidden. Cancellations were in the order history, and driver availability existed in the operation’s own records. What was missing was the connection. The spike didn’t appear as a billing error, because a cancelled order generates no invoice. It didn’t appear clearly in monthly reporting either, because cancellations and capacity were reported separately, and at a level of averaging that smoothed the spike away.

That pattern is typical of what we find. The data to see the leak usually already exists. It is split across systems nobody joins, summarised at a level that hides it, and owned by no one in particular.

How do you find revenue leakage in your own data?

You don’t need new software to get a first read. Start with a quick self-check. If two or more of these apply, it’s worth a closer look:

• Cancellations and failed deliveries are reported monthly, not by week, lane or customer.
• A large share of cancellation reason codes are “other” or blank.
• Demurrage and detention invoices are paid by finance without going back to operations.
• Waiting time and extra stops are agreed by phone or chat and rarely make it onto an invoice.
• Nobody can say which three customers or lanes cost the most in re-work.
• Carrier invoices are approved without checking them against proof of delivery.

Then run a basic analysis with data you already have:

1. Set an expected-revenue baseline per order. For a sample of orders, work out what should have been billed under the contract, including accessorials, and compare it with what was actually invoiced and collected. This catches the billing side.
2. Pull twelve months of order history, including cancellations and failed attempts. Most TMS and order systems can export this directly.
3. Break it down finer than monthly. Look by week, lane, customer and reason code. Spikes that disappear in a monthly average often stand out clearly at weekly level.
4. Join it with one second dataset. Driver or carrier availability, slot bookings, fee invoices or empty-kilometre data are good candidates. Most root causes sit at the join between two systems.
5. Rank what you find by recoverable value. Estimate what each leak costs per month and how much of it is realistically fixable. Work from the top.

This won’t catch everything. A twelve-month export misses what lives in calls and chats, and a manual join gets slow beyond one or two datasets. But it will tell you whether you have an operational leak worth chasing, and roughly where.

Systematic or one-off: what should you fix first?

Not every anomaly is a leak worth fixing. A cancellation spike during a storm week needs no action. A pattern that repeats on the same lane, weekday or customer does.

Before acting on a finding, ask three questions. Does it recur? Is the cause something the operation controls? And is there someone who would own the fix? A leak that recurs, has a controllable cause and has an owner is worth acting on now. One that fails all three is usually noise.

Our view is that most teams try to fix too many leaks at once, and fix the most visible ones rather than the most expensive. Two or three systematic leaks, properly closed, usually matter more than a long list of one-off findings.

Where does Anomaly Detection & Pattern Recognition fit?

The manual method above has limits: it samples, it’s slow to repeat, and it can’t see what happens in calls and chats. Anomaly Detection & Pattern Recognition, one of the solutions that we offer does the same work continuously across an operation’s history.

It surfaces deviations with a named root cause and marks each one as systematic or one-off. It also ranks the repeating manual decisions in an operation by how mechanical they are, which shows where automation would pay off first. It flags; people decide what to do about it.

For operations like those described here, we model up to ~15% of silently leaking revenue recovered, because root causes are found before losses cascade. That is a typical range, not a promise. The actual figure depends on volume, lanes and how the data is split, and can only be measured on the operation itself.

What’s a sensible first step?

Run the self-check and the five steps above. They cost nothing and will tell you whether operational leakage is a real issue for you, and where to look first.

If you want an outside view after that, we run paid pilots on an operation’s own data, with a first result in 7 days and nothing written to live systems. Whether or not that’s the right next step, the self-check is.

If your leaks trace back to an overloaded dispatch desk, our guide on how to reduce dispatcher workload covers that side. For the wider picture, see AI in logistics: what it does and where it pays off.

FAQ

What is the difference between billing leakage and operational leakage?

Billing leakage is unbilled or uncollected revenue for transport services already executed, whereas operational leakage is unearned revenue or avoidable operational expense caused by execution failures such as cancelled shipments or unproductive empty mileage.

Why do carrier invoice overpayments occur?

Carrier invoice overpayments occur when transport invoices are processed without systematic three-way reconciliation against purchase order commitments and verified proof of delivery records.

How does operational latency increase days sales outstanding?

When shipping documents, delivery receipts, and accessorial approvals are assembled manually, invoice creation is delayed three to five days, which slows cash collection across completed freight movements.

Sources

• American Transportation Research Institute (ATRI) — https://truckingresearch.org/2024/09/new-research-documents-substantial-financial-and-safety-impacts-from-truck-driver-detention/
• Eurostat’s road freight statistics — https://ec.europa.eu/eurostat/statistics-explained/index.php/Road_freight_transport_by_journey_characteristics
• How to Reduce Dispatcher Workload — https://blog.platform24.ai/blog/how-to-reduce-dispatcher-workload/
• AI in Logistics: What It Does and Where it Pays Off — https://blog.platform24.ai/blog/ai-in-logistics-what-it-does-and-where-it-pays-off/