- What does firing a customer actually mean?
- Which accounts are actually candidates?
- What goes wrong when you just cut them off?
- From diagnosis to action
- The sequence, in order
- What if the account you want is on the other end of the same contract?
The question people ask is how to fire an unprofitable customer. The question they actually need answered is narrower and harder: how do I stop losing money on this relationship without damaging the three other relationships attached to it that nobody has mapped.
That second version is the one worth writing about, because the first version has an easy answer that is usually wrong. If you have already built the customer and product profitability picture and you are staring at a quadrant of accounts that lose money, the instinct is to send letters. Hold that instinct for about two more weeks.
What does firing a customer actually mean?
In practice it almost never means ending the relationship, and treating it as a binary is how the whole exercise goes sideways. What you are firing is a set of terms: a price, a lead time, a minimum order quantity, a freight arrangement, a support level, a payment window. The customer is the counterparty to those terms, not the terms themselves.
I have watched several of these go badly and the pattern is consistent. The account gets identified on a spreadsheet, a decision gets made at a level that has never spoken to anyone at the customer, and a letter goes out. What arrives back is not a quiet exit. It is a call from somebody two levels above the person you wrote to, at a parent company you did not realize was a parent company.
You are firing a set of terms, not a counterparty. Confusing the two is how a $40,000 problem becomes a $2M conversation with somebody’s corporate office.
Which accounts are actually candidates?
Far fewer than your first list suggests, because most of the first list is an allocation artifact rather than a real loss. Before an account is a candidate it has to survive three tests, and in my experience roughly half of any initial list does not.
The first test is whether the loss survives an honest cost allocation. If your overhead is spread on revenue or on direct labor, small accounts will look unprofitable whether they are or not, because that method assigns them cost in proportion to the wrong thing. Reallocate on the activities that actually drive the cost: order lines, setups, expedites, engineering hours, returns.
The second test is whether the loss is structural or self-inflicted. An account that loses money because you granted 2.5 percent extra discount in a soft quarter four years ago is not an unprofitable customer. It is an unprofitable decision that has been renewing itself automatically ever since, and it is fixable without touching the relationship.
The third test is whether the account is small on its own or small inside something larger. This is the one people skip, and it is the one that hurts.
What goes wrong when you just cut them off?
Three failure modes, and all three are avoidable with a week of work that nobody wants to do because it is unglamorous and involves asking your own sales team questions.
The first is the hidden parent. Small buying entity, large corporate group, shared procurement function, and a category manager who now has your name in a memo. This happens most often with regional operations of national accounts and with divisions bought by a group in the last two years, where your customer master has never been updated to reflect who owns whom.
The second is the mobile buyer. The purchasing manager at the account you fire this year is the purchasing manager at a much better account in three years, and people remember the exit conversation far longer than they remember the price. This is not sentiment. It is a real, repeated, and entirely predictable cost that never appears on the spreadsheet that justified the decision.
The third is the capacity that never gets filled. The financial case for exiting an account almost always assumes the freed capacity goes somewhere. If it does not, you have removed contribution and kept the fixed cost, and your margin percentage improved while your actual dollars got worse. That is the counterexample that should make you slow down: the exercise can be executed perfectly and still leave you poorer.
Margin percentage improved and margin dollars fell. That is what a technically successful customer exit looks like when the freed capacity had nowhere to go.
From diagnosis to action
Everything above is about deciding whether this account belongs on the list at all. What follows assumes it survived all three tests, the cost is real, the loss is structural, and you have mapped who owns whom. Now the sequence matters more than the decision.
The sequence, in order
Run these in order and most of the accounts on your list will resolve themselves before you reach step six. That is the intended outcome, not a failure of nerve.
- Confirm the loss with activity-based cost, not allocated overhead. Rough is fine. Three weeks of approximate allocation beats six months of precise allocation you never finish.
- Map the ownership graph before you map the P&L. Who owns this entity, who else do they own, and does any of it buy from you under a different name. Your ERP will not tell you this. Your salespeople will.
- Identify which specific cost is doing the damage. It is usually one thing, not the account in general: expedite frequency, order line size, a custom variant, returns, or payment terms. Name it before you price it.
- Change the terms, not the relationship. Build the offer around the specific cost driver. A minimum order quantity, a standard lead time, a small-order fee, a move to the standard product, or a price that covers the variant. Keep everything else identical.
- Put a number and a date on it, in writing, with a person’s name attached. Vague repricing gets negotiated into nothing. Specific repricing with an effective date gets a yes or a no, and either is useful.
- Hold the price through the first pushback. There is always a first pushback and it is almost always reflexive. The accounts that were genuinely going to leave leave at the second one, not the first.
- If they go, exit on terms. Notice period, last-time buy, a named alternative supplier, and a call rather than a letter. You are managing a reference you will never control again, and the cost of doing this well is one afternoon.
Steps one through four are where the money is. In the exercises I have run, a clear majority of flagged accounts either turn profitable on new terms or were never unprofitable once the cost was allocated properly. The letters at step seven are the rare case, which is the opposite of how this topic is usually written about.
What if the account you want is on the other end of the same contract?
Then you do not touch the contract, you change what sits inside it, and you tell the person who runs the relationship before you tell anyone else. This is the hard version and it is more common than the clean version, particularly in multi-site and multi-division supply agreements.
The move is to separate the profitable and unprofitable activity inside the existing commercial frame. The custom variant gets retired at the next revision. The low-volume ship-to sites move to a distributor or to a minimum order quantity. The expedite provision gets a fee attached at renewal. None of that requires renegotiating the agreement, and none of it produces the phone call you are trying to avoid.
The one rule I would not break: the person who owns the large relationship hears about it from you, in advance, with the arithmetic, before any customer contact happens. They will usually agree. What they will not forgive is finding out from their customer, and if that happens you have traded a margin problem for an internal one, which is a worse trade than it sounds.
Tell the person who owns the relationship first. They will usually agree. What they will not forgive is hearing it from their own customer.
Monday morning: pull your flagged account list and add one column for parent company, filled in by the salesperson who owns each account rather than by finance.
About the Stagnation Assassin
Todd Hagopian is a Fortune 500 transformation executive who has generated $3B in shareholder value across Berkshire Hathaway, Illinois Tool Works, Whirlpool, and JBT Marel, where he serves as Global VP of Product Strategy and Marketing. Known as The Stagnation Assassin, he is the author of two published books: The Unfair Advantage: Weaponizing the Hypomanic Toolbox and Stagnation Assassin: The Anti-Consultant Manifesto. His blog is published in 15+ languages and read by operators worldwide. Bring him to your stage via his speaking page or connect with him on LinkedIn.
Want to know whether your customer base is the cause or the symptom? Stagnation Genome scoring reads where a business has stopped moving and why, and a tail of accounts nobody will reprice is one of the clearest markers it picks up. Request a scoring session here.