How do you find customers who pay too little?

You find customers who pay too little by comparing, for each customer, the unit price actually invoiced with what they should pay under your pricing policy and their own agreements. The customers furthest below that norm, with no recorded reason, are your leak. It is usually a matter of discounts that were never reversed, agreements that have expired, wrong volume tiers or services that are delivered but do not appear on the invoice.

What does paying too little actually mean?

Paying too little is not the same as a low price. A large customer with a sharp price that was deliberately negotiated is not paying too little. They are paying what was agreed. The leak sits in three other situations:

1. The customer pays less than was agreed. The contract says EUR 95 an hour, the invoice says EUR 85. Or the contract includes indexation that was never applied.
2. The customer pays less than your policy allows, without anyone signing off on it. An account manager gave 25 percent off when the limit was 15 percent, and nobody noticed.
3. The customer gets more than they pay for. They use more licences, more hours, more volume or a bigger package than appears on the invoice.

The first is an execution error. The second is a policy error. The third is a delivery error. Each needs a different check, but the result is the same: revenue you have earned and do not receive.

Where does it come from?

In B2B the same patterns come up again and again:

• A temporary discount that became permanent. "Three months at 20 percent off to get started." The discount is set up as a standing line in the billing system, with no end date. Two years later it is still running.
• Volume tiers that do not move. The customer got a volume discount at 500 units a month. They now buy 200, but the tier price was entered as a fixed customer price and stays in place.
• Old price agreements in the ERP. In Exact, AFAS, SAP, NetSuite or Dynamics the customer has its own price that overrides the price list. You raise the price list, but that customer does not notice a thing.
• Additional work and extras that are not charged. Express deliveries, extra meetings, changes outside the scope. They are done, but nobody creates an invoice line.
• Licences or users above the agreed numbers. In software and managed services, usage grows while the invoice stays at the number from the start.
• Missed price increases. A category of its own, covered separately in how to find missed price increases.

Method: the price distribution per product

The fastest way to see who pays too little is to look at the spread in realised price for each product or service. All you need is invoice data.

1. Export all invoice lines from the past twelve months. Customer, item or service, quantity, amount, discount.
2. Calculate the net unit price for each line. Amount after discount divided by quantity.
3. Group by product. For each product, calculate the list price, the median of what customers actually pay, and the lowest prices.
4. Flag customers well below the median. A threshold of 15 to 20 percent below the median is a workable start. Adjust it if the list is too long or too short.
5. Add the volume. A customer 40 percent below the median who buys once a year is less urgent than a customer 10 percent below the median who buys every month.
6. Find the reason for each flagged customer. Contract, written agreement, approval. If there is one, record it in the system. If there is not, you have found a leak.

This works in Excel or Power BI. It needs no integrations, only a good export. The drawback is that you only see what was invoiced. Whatever never made it onto an invoice stays invisible.

Method: agreement against invoice

For the first and third situations, less than agreed and more delivered than paid for, you need a second source next to the invoice:

• The contract or order confirmation for what was agreed. The way you check sales orders against invoices applies directly here.
• Time tracking, the delivery system or usage data for what was actually delivered.

Put them next to the invoice for each customer. Every difference not offset by a credit note or an approved deviation is a candidate.

Worked example

Worked example: suppose you have 300 active customers and invoice EUR 8 million a year. The price distribution analysis shows 30 customers sitting on average 12 percent below the median price with no recorded reason. Together they buy EUR 1.2 million.

• What those 30 customers would pay at the median: EUR 1.2 million divided by 0.88, rounded to EUR 1,364,000.
• The difference: roughly EUR 164,000 a year.

You will not recover all of it. Some customers will leave if you correct the price, and with others you will deliberately choose not to strain the relationship. But even if you partly correct half of them, the return is greater than the work of the analysis.

This is an example, not an average. How much it is for you depends on how much freedom account managers have, how old your customer base is and whether you record price agreements with an end date.

Checklist for tomorrow

• Are all discounts in the billing system a separate line, or are they built into a customer price? A separate line can be checked, an adjusted price cannot.
• Does every discount have an end date or an explicit note saying "permanent"?
• Is there a limit for discounts without approval, and is it enforced in the system?
• When were the customer-specific prices in the ERP last reviewed?
• Is additional work invoiced from time tracking or from the project manager's memory?
• Is the number of licences, users or units on the invoice periodically compared with actual usage?

Every no on this list is a place where customers can pay too little without anyone seeing it. It fits into the broader approach to finding revenue leakage, where pricing is one of the standard areas.

How to have the conversation with the customer

A customer who has paid a price that is too low for years regards that price as agreed. The fact that they are formally paying too little does not make the conversation easier. A few principles:

• Start with the agreement. Where the contract specifies a higher price or indexation, refer to it. That is a correction, not a negotiation.
• Correct going forward, not backward. Recovering the past is usually not worth it legally or commercially. The gain lies in the years ahead.
• Choose the moment. A renewal, an expansion or the annual price adjustment are natural moments.
• Phase it if necessary. A 20 percent correction in one go is difficult. Two steps over twelve months is often acceptable.

Why you do not see this in a CRM report

The CRM shows what was sold, often including the deal value at closing. It does not show what is actually invoiced twelve months later, nor what is delivered. A customer who pays too little looks like a healthy, loyal customer in the CRM.

That is why this is a check across systems. The example patterns on the use cases page show how discounts and prices that never moved give themselves away in the data. The causes are covered further in revenue leakage from wrong prices and, for recurring revenue, in revenue leakage from wrong subscriptions.

Frequently asked questions

Is a customer with a large discount a problem by definition?

No. A large discount given deliberately, with approval and a reason, is a commercial choice. The problem is a discount with no trail: nobody remembers why, nobody signed off and there is no end date.

How often should I review the price distribution?

Every quarter is enough to spot new deviations in time. The first time takes a few days. After that, with the same export and the same thresholds, it goes quickly.

What if I do not have proper invoice lines, only totals?

Then that is the first finding. Without lines per product or service you cannot compare prices. Start with your largest customers and pull up the underlying orders.

Who should do this analysis?

Finance has the data, sales knows the reasons. The analysis belongs with finance, the assessment per customer with the account owner, and the decision on corrections with the leadership team. Without an owner, the list goes nowhere.