How do you find revenue leakage in a business?

You find revenue leakage by comparing, at each step of your revenue chain, what should have happened with what did happen. Won deals against invoices, contracts against invoice lines, work delivered against work invoiced, contract value against realised revenue. Every difference without a recorded reason is a possible leak. Start at the handover with the highest volume and the most manual work, convert every difference into euros per year, and then make the check recurring rather than one-off.

Why you do not see revenue leakage on its own

Revenue leakage is revenue you were entitled to and did not receive. An indexation that was not applied, a job sheet that was not invoiced, a customer who uses more than they pay for. What all of these have in common: no alarm goes off. Your receivables ledger is clean, your invoices are correctly formatted, your customers are happy. The money that is missing is not recorded anywhere as missing.

That is the fundamental problem. Financial reporting shows what happened. Revenue leakage is what did not happen. You only find it by setting an expectation beside reality. What leakage is exactly, and the forms it takes, is covered in the complete guide to revenue leakage. This article is about how to find it.

How do you know if revenue is slipping away?

There are signs that show something is wrong without any in-depth analysis. They do not prove a leak, but they tell you where to look.

• Sales and finance quote different revenue figures for the same period, and nobody can explain the difference exactly.
• The margin falls while revenue rises, without costs having risen proportionately.
• Invoicing depends on one person who knows which customers have exceptions.
• Credit notes are an everyday occurrence, because invoices have to be corrected after the event.
• Prices have not been adjusted for years even though the contracts allow indexation.
• Customers never ask for a correction in your favour. If customers complain about invoices that are too high but never about invoices that are too low, that tells you nothing about how often too-low invoices occur.

A longer list, with what each sign usually points to, is in 10 signs your business is leaving revenue on the table. If you recognise three or more, the question is not whether there is leakage, but how much and where.

Three ways to search

There are three ways to track down revenue leakage. They complement each other.

Top-down: expectation against reality. You calculate what revenue should have been on the basis of contracts, orders and rates, and compare that with what was invoiced. The difference is the total leakage. This tells you how much, but not where. How to calculate that difference is covered in how to calculate revenue leakage.

Bottom-up: a check at each handover. You walk the revenue chain step by step and check at every handover whether everything that goes in also comes out. From won deal to contract, from contract to order, from order to invoice. This tells you exactly where it leaks, and it is the core of this article.

Patterns: deviations between customers. You compare customers with each other. Customers who pay much less than average for the same product, customers whose volume is falling without anyone knowing, customers who use more than comparable customers but pay the same. This finds leaks that break no rule at all.

In practice you start bottom-up, because that produces concrete findings fastest. You use top-down to check whether you have found everything. Patterns are the last layer, and that is where automation and AI come into their own.

The revenue chain: where to look

Every B2B company has a revenue chain, even if nobody has ever drawn it. From first contact to paid invoice and renewal. At every transition between two steps, something can be lost.

For each transition there is an article that works through the check step by step.

Is all revenue being invoiced? The broadest check: has everything that was delivered or sold actually reached an invoice? It starts with establishing what "everything" is, and that is harder than it sounds. See how to check that all revenue is invoiced.

Are contracts billed correctly? This is not about whether there is an invoice, but whether the invoice matches the agreements: rates, discounts, indexation, scope, invoicing frequency. See how to check that contracts are billed correctly.

Does the CRM agree with billing? The CRM says what was sold, billing says what was invoiced. The two should match, and rarely do exactly. See how to check CRM against billing.

Do sales orders match invoices? For companies that work with orders, such as wholesalers, manufacturers and installers, the order is the source. Partial deliveries, back orders and changes after the order are the classic places where lines disappear. See how to check sales orders against invoices.

Does a contract deliver what it promised? A contract worth EUR 120,000 over three years should yield EUR 120,000, plus indexation. If after three years it has yielded EUR 104,000, you want to know why. See how to check contract value against realised revenue.

Have invoices been forgotten? Sometimes there is no error in an invoice; the invoice simply does not exist. Work that was carried out but never invoiced, or a periodic invoice that was not created one month. See how to find forgotten invoices.

How do I know where my business is leaving revenue on the table?

Besides the checks per transition, there are five specific leaks that occur in almost every B2B company. They are worth looking for because they are often large and can be found with a targeted question.

Missed price increases. Contracts with an indexation clause that was not applied, or price lists that were not updated. The leak compounds, because every missed increase is the base for the next. See how to find missed price increases.

Customers who pay too little. Customers who consistently pay less than comparable customers for the same product or service. Sometimes there is a reason, often it is a discount that was once given and never withdrawn. See how to find customers who pay too little.

Missed upsell. Customers who use more than they pay for, or who are ready for a larger package without anyone asking. See how to find missed upsell.

Customers quietly buying less. No cancellation, no complaint, but volume falls month after month. By the time anyone notices, the customer has often largely gone. See how to spot customers who are quietly buying less.

Errors in recurring revenue. Subscriptions, maintenance contracts and retainers that do not match what was agreed or what is being used. Because they repeat every month, the error repeats too. See how to find errors in recurring revenue.

Worked example: what a first round turns up

Worked example: suppose a technical services company with EUR 6 million in revenue, 220 maintenance contracts and its own service department carries out a first round of checks over the past year. It looks at three transitions and two specific leaks.

That is just over 2 percent of revenue. It falls within the range of 1 to 5 percent commonly cited as an estimate for revenue leakage, but the percentage is not the point. The point is that each of these amounts has a name, a customer and a cause. You can do something with that. You cannot with a percentage.

Note also what is not in this overview: the customers who have quietly started buying less and the upsell nobody proposed. You do not find those with a check per transition, but with pattern recognition. They come in a second round.

Manual or automatic

A first round is almost always done by hand: exports from the CRM, the accounting package and time tracking, combined in a spreadsheet. That is fine. It forces you to understand how your data fits together, and it produces concrete findings within a few weeks.

But manual checks have a shelf-life problem. They take a lot of time, so they are done once a year, and new leaks arise in the meantime. A job sheet left lying in February is only found in December, when the customer no longer remembers the work.

Automatic detection solves that by running the same comparisons continuously. Every night, won deals are set beside invoices, job sheets beside invoice lines, contracts beside prices. Deviations arrive as an alert with an owner, not as a report at the end of the year. How to set that up, with which rules and which thresholds, is covered in how to detect revenue leakage automatically.

The difference is not only speed. An automatic check is not affected by the person who knows how things work being on holiday. And it does not look selectively at the customers who happen to stand out.

What AI can and cannot do

AI is increasingly put forward as the answer to revenue checking. That is partly justified. But it helps to know where AI adds something to rule-based checks and where it does not.

What rules do well. A won deal without an invoice, an invoice price that deviates from the contract price, a job sheet without an invoice line. These are comparisons with a clear right and wrong. You do not need AI for them. A well-designed rule is faster, cheaper and easier to explain.

What AI adds. AI helps where there is no fixed rule, or where the information is not in tidy fields.

• Reading unstructured sources. An indexation clause on page seven of a PDF contract, an agreement on additional work in an email, an unusual payment arrangement in a note. AI can read that text and turn it into fields you can compare. How that works for contracts is covered in can AI check whether contracts are billed correctly.
• Linking records that do not match neatly. "Baker Ltd" in the CRM and "Baker Installation Services Ltd" in the accounts. A deal of EUR 48,000 and three invoices of EUR 16,000. AI can propose likely links where an exact comparison fails. See can AI see when CRM and billing do not match.
• Recognising unusual patterns. A customer whose volume deviates from their own history or from comparable customers, without a rule that says what normal is. This is where anomaly detection and predictive models earn their keep. See can AI detect revenue leakage, including the question of how exactly AI detects a revenue leak.
• Looking ahead. A customer whose behaviour resembles that of customers who left last year, or a contract whose structure makes it likely to leak. That is prediction, and there the limits matter more than the possibilities. See can AI predict revenue leakage.

What AI cannot do. AI cannot find data that does not exist. A verbal agreement that was never recorded does not exist for AI. Nor can AI decide whether a deviation is a leak or a deliberate choice. A 30 percent discount can be a mistake or a strategic arrangement with a key customer. That judgement stays with a person. And AI is only as good as the data it is given: with duplicate customers and empty fields, AI produces just as many false alerts as a rule.

Do it yourself, have it done, or monitor continuously

There are three routes to finding revenue leakage. They are not mutually exclusive.

Do it yourself. With the steps in this article, your own exports and a spreadsheet. This works well if you have someone with the time, access to all systems and the skills to link data. The drawback: it often stops after one round, and your organisation's blind spots are also the blind spots of your own check.

A one-off review. An outside party walks through your revenue chain and prices the findings. AutoMaat does this with the Revenue Audit: a one-off, personally conducted review of eight areas (pipeline, conversion, pricing, invoicing, retention, support, marketing and data quality), in five blocks of questions that adapt to each company. Every finding is given an amount in euros. The audit costs EUR 1,500, with the guarantee that you do not pay if it finds less leakage than it costs. Nothing has to be connected, and the plan that comes out of it is yours.

Continuous monitoring. A platform that connects to your existing systems and runs the comparisons continuously. RiOS is such a platform, currently in beta. It is a separate product from the audit: it connects to the systems you already use, such as CRM, billing and support, monitors revenue leakage continuously and is priced per company, not per user. You can use one without the other.

The right route depends on where you are. If you do not yet know whether and where you are leaking, start with a first round, yourself or with outside help. If you know you are leaking and want to stop it coming back, the question is how to make the check recurring.

Step-by-step plan: the first 30 days

You can start this plan tomorrow, with the systems you already have.

Week 1: map it.

1. Draw your revenue chain: from first contact to paid invoice and renewal. Which systems are used at each step?
2. Mark every handover between systems or between people. Where is something retyped, forwarded by email or edited in a spreadsheet?
3. Choose the three handovers with the highest volume and the most manual work. That is where you start.

Week 2: collect the data.

4. For those three handovers, export the data from both sides for the past twelve months. For example: won deals and invoices, job sheets and invoice lines, contracts and subscriptions.
5. Build a mapping table for customers across systems. This is often the most work, and you will need it again for every subsequent check.

Week 3: compare.

6. Set the data side by side. Mark every item that appears on one side but not the other, or with a different amount.
7. Investigate the twenty largest differences. Is there an explanation? Has it been recorded?
8. Convert every confirmed leak into euros per year.

Week 4: decide and embed.

9. Decide for each finding what you will do: invoice it, correct it from now on, or deliberately leave it with a recorded reason.
10. Appoint an owner for each check and agree a frequency. A check without an owner has disappeared within three months.
11. Choose the next three handovers for the second round.

What not to do

• Do not start with a new system. You find the first leaks with what you have. A new system only helps once you know what you want to compare.
• Do not do everything at once. A check across every handover at the same time gets bogged down in data work. Three handovers done well yield more than ten done by halves.
• Do not stop after one round. Most leaks reappear as soon as attention slips. A check that does not recur is a snapshot.
• Do not treat every deviation as an error. Some differences have a good reason. The goal is not zero deviations, but zero unexplained deviations.

Frequently asked questions

How long does it take to find revenue leakage?

A first round across three handovers takes about a month with the systems you already have, depending on how well your data lines up. You usually have the first concrete findings within two weeks.

Where do I usually find the largest leaks?

That differs by type of business. For service providers it is often in unbilled hours and additional work. For companies with maintenance contracts, in indexation and expired contracts. For subscription businesses, in wrong quantities and packages. Start at the handover with the most volume and manual work.

How much revenue leakage is normal?

A commonly cited estimate is 1 to 5 percent of revenue. For your company that says little. The only way to know the real figure is to measure it: per handover, per finding, in euros.

Can I find revenue leakage without connecting my systems?

Yes. With exports from your systems and a spreadsheet you can get a long way. Connecting only becomes important when you want to make the check recurring without it costing days of manual work each time.

Do I always have to recover leakage I find from the customer?

No. What you do about the past is a commercial and sometimes legal judgement for each customer. What you always do: correct it from now on, and record what you decided and why.