Every company has them. A handful of customers who buy regularly, pay their invoices, appear on the sales team’s list of wins, and quietly destroy value on every order. They are not obvious, because nothing in a standard profit and loss statement is designed to expose them. Revenue looks healthy. Gross margin looks acceptable. The loss is hiding one layer below, in the cost of serving them.
How do you identify unprofitable customers?
Rank every customer by profit after cost to serve, not by revenue or gross margin. Assign the cost of the activities each customer actually consumes, such as orders, deliveries, returns, support calls and credit, then sort the list from best to worst. Customers whose consumed cost exceeds their contribution appear immediately, and they are usually not the ones you expected.
Your biggest customers are the usual suspects
The reflex is to assume that small accounts are the problem. Most of the time it is the opposite. Large customers negotiate the deepest discounts, demand the shortest lead times, split orders across more deliveries, request custom packaging or reporting, and take longer to pay. Every one of those concessions consumes resources that never appear on their invoice.
The result is a familiar shape. When you rank customers by cumulative profit, the curve climbs steeply, flattens, and then falls. That decline is real money being handed back, and the customers responsible for it are almost always concentrated at the top of the revenue list. This is the pattern behind the whale curve, and it is the fastest way to make the problem visible to a management team that does not want to believe it.
Gross margin will not find them
Gross margin stops at the cost of the product. It says nothing about how much work it took to sell, deliver, correct and collect. Two customers buying the identical product at the identical price can differ by twenty points of real margin once you count what happened after the sale.
This is why an average is dangerous. A blended margin across a customer base flatters the losers and penalises the winners, and it produces exactly one decision: keep doing what you are doing. If the figure you review every month is an average, you have no way of knowing which customers are subsidising which.
Start with the activities, not the accounts
The instinct is to open the ledger and try to split each cost line across customers. That approach stalls quickly, because a general ledger is organised by what you bought, not by what you did. Start instead with the work.
List the activities that consume capacity outside production: taking an order, picking it, delivering it, handling a return, answering a support request, resolving a dispute, chasing payment. For each one, estimate how long it typically takes and how often each customer triggers it. You do not need a time and motion study. A credible estimate from the people doing the work is enough to change the ranking, and the ranking is what you are after. This is the core of cost to serve.
Build the ranking and read it from both ends
Once each customer carries the cost of the activities they actually consume, sort them by profit. Read the list twice.
From the bottom, you find the accounts that are losing money and why. Usually the reason is one specific behaviour: a customer who orders in very small quantities, or returns a fifth of what they buy, or absorbs disproportionate technical support. From the top, you find something more useful. You discover which customers are cheap to serve and profitable at a lower headline price, which tells you exactly what kind of business to go and win more of. A step by step customer profitability calculation gives you both ends at once.
An unprofitable customer is rarely a bad customer. It is usually a good customer on the wrong terms.
What to do once you can see the list
Firing customers is the loudest option and almost never the first one. The behaviour that creates the loss is usually negotiable. Consolidate the small orders into a scheduled delivery. Set a minimum order value. Move a low-margin, high-touch account onto self-service ordering. Charge for the expedited shipping that was previously absorbed. Reprice at renewal, with the cost evidence in hand.
What changes is not the customer relationship. It is the fact that you are now negotiating with the numbers in front of you, instead of guessing which concessions you can afford. Most accounts at the bottom of the list can be moved back above the line without losing them, and the ones that cannot are now a deliberate choice rather than an accident.
A structured Profitability Health Check shows you which customers, products and channels are carrying the business, and which are quietly draining it.