When Should You Fire a Customer? A Profitability and Fit Test

The hardest customer to let go of is often not the smallest one.

It may be the customer with the biggest logo, the longest relationship, the most revenue, or the account the founder personally fought to win.

That is exactly why the decision can become distorted.

A customer can produce substantial revenue while consuming disproportionate support time, custom work, discounts, management attention, collection effort or product capacity. Research on customer profitability has repeatedly shown that customer size by itself does not determine customer profitability. The economics depend on what it costs to acquire, serve and retain the relationship, not simply what the customer pays.

So the question is not:

"Is this customer difficult?"

The better question is:

"Can this customer relationship work inside the business we are actually trying to build?"

If the answer remains no after reasonable attempts to change the economics or operating model, ending the relationship can be a rational business decision.

Fire a customer when the relationship is persistently uneconomic, operationally destructive or strategically misaligned, and you have already tested reasonable ways to repair the fit.

That definition has an important safeguard.

Do not use "bad customer" as a label for a customer who asks hard questions, negotiates, complains about a real problem, or expects the service they were promised.

The customer may be difficult because the business made a bad promise.

The customer may be expensive because the pricing model is wrong.

The customer may be demanding because the product genuinely does not work.

Those are business problems to diagnose before the relationship is terminated.

Customer profitability research provides a useful correction to a common founder mistake: treating revenue as the same thing as value.

A 2005 study of customer profitability found that the relationship between customer size and profitability is not direct. Other variables, including exchange efficiency and the economics of serving the account, matter. Earlier work on customer profitability similarly argues that profitability should be analyzed at the customer level rather than inferred from broad averages.

In practical terms, two customers paying the same amount can create very different economics.

Customer A pays $5,000 a month.

Customer B also pays $5,000.

But Customer A:

• uses the standard service;
• pays on time;
• needs little support;
• accepts normal delivery terms;
• renews without heavy negotiation.

Customer B:

• requests frequent custom work;
• needs repeated revisions;
• requires senior staff in routine meetings;
• receives special pricing;
• pays late;
• interrupts planned work;
• expects exceptions that were never priced.

The revenue report says the accounts are equal.

The operating system says they are not.

That difference is where the decision begins.

Source: https://doi.org/10.1108/08858620510592768

Source: https://doi.org/10.1002/(SICI)1520-6656(199924)13:1 3.0.CO;2-L

This is one of the most important distinctions.

A customer can be emotionally difficult and economically excellent.

Another customer can be pleasant and quietly unprofitable.

Founders often overweight the first category because emotional friction is visible while cost-to-serve is hidden.

Before ending an account, ask:

1. Does the customer make a reasonable economic contribution?
2. Are the demands inside the contract or offer?
3. Is the business repeatedly providing unpaid work?
4. Does the account require scarce senior capacity?
5. Does the customer delay payments or create collection work?
6. Are discounts permanent rather than temporary?
7. Is the account forcing product decisions that hurt the broader customer base?
8. Is the relationship strategically useful even if current margin is modest?
9. Could the problem be fixed by changing price, scope, service level or expectations?
10. Have we actually tried the correction?

A "yes" to several of these does not automatically mean "fire them."

It means the relationship deserves an economic diagnosis.

Ending the relationship should usually be the fifth option, not the first.

If the customer is profitable only under an unrealistic level of founder involvement, the price may be wrong.

A customer who needs twice the service of the standard account may need a different price.

Do not apologize for pricing that reflects real delivery costs.

The important test is whether the customer accepts the economics.

If the answer is no, you have learned something about the fit.

Some "difficult customers" are actually customers operating inside an undefined scope.

If the offer says "ongoing support" without defining what support means, almost any support request can become a dispute.

Define:

• what is included;
• what is excluded;
• response expectations;
• revision limits;
• meeting limits;
• turnaround;
• custom work;
• emergency work;
• additional fees.

A customer who becomes healthy after the boundary is clarified may not be a bad customer at all.

Some accounts are expensive because the company serves them manually.

A different onboarding process, support tier, communication cadence or delivery model can change the economics.

The customer may be worth keeping if the service architecture changes.

Sometimes the customer is exposing a genuine product problem.

If several customers make the same complaint, the answer is not to fire them.

The complaint may be evidence that the offer is broken.

Customer intervention research argues that businesses should consider separating segments, changing prices or service arrangements, and redesigning the relationship before deciding whether termination is appropriate.

Source: https://doi.org/10.1080/15332660802508505

If the relationship remains structurally wrong after reasonable corrections, ending it becomes the cleanest option.

That does not require anger.

It requires clarity.

Profit is not the only dimension.

Consider an account that produces a healthy margin but repeatedly pushes the company toward a business it does not want.

For example:

A software company builds a focused product for a defined customer segment.

One large customer pays well but repeatedly demands custom features for its unusual internal workflow.

The founder keeps saying yes because the account is profitable.

Over time, the roadmap becomes a collection of exceptions.

New customers see a more complicated product.

The team supports edge cases.

The sales message becomes harder to explain.

The product is technically stronger for one account and strategically weaker for everyone else.

This is not simply a customer-profitability problem.

It is a strategy problem.

Your existing SOE guidance on custom work and customer requests addresses related decisions, but the customer-level question is different: does this particular relationship belong inside the operating model?

That distinction matters because a company can have profitable customers who are still poor strategic fits.

You do not need a complex financial model.

Start with six dimensions and score each from 1 to 5.

Do not average the score blindly.

A customer with strong economics but disastrous strategic fit may still deserve a different decision.

The purpose of the scorecard is to expose the trade-offs.

One of the most useful lessons from customer profitability research is that the relationship's economics depend on the interaction between customer characteristics and the firm's own operating model.

A demanding customer is not necessarily objectively unprofitable.

The same customer may be profitable for one company and unprofitable for another.

Why?

Because the companies have different:

• pricing;
• staffing;
• delivery costs;
• service promises;
• automation;
• product architecture;
• customer expectations;
• cost structures.

That means "fire the customer" is not a universal rule.

The right question is:

"Does this customer fit our economic model?"

A customer relationship deserves serious review when several of these patterns persist.

The account looks better in the sales dashboard while the team feels increasingly overloaded.

That is a classic signal to examine cost-to-serve.

If every renewal creates a new exception, the company may be selling a custom business disguised as a standard product.

A customer that requires the founder or senior team for work that should be standardized can create an invisible opportunity cost.

Late payments are not just a finance problem.

They create forecasting uncertainty, collection work and sometimes financing costs.

If one account repeatedly forces product decisions that make the core product worse for the target market, the account has strategic influence beyond its revenue.

A relationship can be expensive even when the direct cost is difficult to measure.

Repeated escalation, negotiation and exception handling consume attention.

If you decide the relationship should end, do it professionally.

First, review the contract and obligations.

Second, document what is being delivered and what remains.

Third, decide the transition period.

Fourth, communicate clearly without blaming the customer.

Fifth, protect customer data and access.

Sixth, complete the agreed handoff.

Seventh, record what the relationship taught the business.

Do not turn the termination into a public story about how terrible the customer was.

The lesson belongs inside the operating system, not in a revenge post.

Jason Fried and 37signals have written publicly about avoiding customers who are not the right fit as part of avoiding unnecessary complexity and overkill. That is useful founder evidence, but it should not be treated as proof that firing customers always improves performance. It is an example of an operating philosophy: a durable company can define what it will not take on.

Source: https://basecamp.com/overkill

The broader lesson is more important than the example.

A business creates value partly by deciding what it will not serve.

There are several situations where termination is premature.

Fix the defect.

Deliver it or renegotiate honestly.

A new customer may require onboarding investment before becoming efficient.

Look at the expected relationship, not one month.

If the account provides meaningful learning, credibility or access to a target market, a lower short-term margin may be rational.

Make the trade-off explicit.

Do not make a major customer decision from frustration alone.

If the account is questionable but the decision is not urgent, run a short structured test.

Record:

• revenue;
• direct delivery cost;
• support time;
• founder/senior time;
• discounts;
• custom work;
• payment delays;
• exceptions.

Tell the customer what will change.

Reprice, narrow the scope, change service levels or stop unpaid exceptions.

Do not judge the customer's reaction to the conversation alone.

Look at behavior.

Do they accept the new economics?

Do they follow the new process?

Do they continue requesting exceptions?

Choose one:

• keep;
• keep with revised terms;
• move to a different service tier;
• pause;
• end.

The value of this process is that it converts emotion into evidence.

The best customer relationship is not necessarily the biggest one.

It is the one that works.

It works economically.

It works operationally.

It works strategically.

It works for the customer.

And it works without requiring the business to become something it does not want to be.

That is why customer termination should not be framed as "getting rid of difficult people."

It is a business-design decision.

Sometimes the right answer is to fix the offer.

Sometimes it is to change the price.

Sometimes it is to change the service model.

And sometimes the honest answer is that the relationship no longer fits.

When that happens, ending it is not failure.

It is boundary-setting at the business level.

Take your ten largest or most time-consuming customers.

For each, estimate:

Customer contribution = customer revenue − direct cost − identifiable service cost − meaningful account-management cost.

Then add the non-financial questions:

• Does this customer fit the market we want?
• Does the account create reusable learning?
• Does it make the product better or more complicated?
• Does the relationship create leverage or distraction?
• Would we choose this customer again today?

The last question is often the most revealing.

A business should not keep a relationship simply because it once worked.

It should keep it because it still makes sense.

This article synthesizes research on customer profitability and customer intervention with public founder/operator material. The research does not establish a universal threshold at which a company should terminate a customer. The decision depends on the firm's economics, service model, strategic priorities and contractual context.

Key sources:

• Van Triest, S. (2005), "Customer size and customer profitability in non-contractual relationships," Journal of Business & Industrial Marketing. https://doi.org/10.1108/08858620510592768
• Mulhern, F. J. (1999/2000), "Customer Profitability Analysis: Measurement, Concentration, and Research Directions," Journal of Interactive Marketing. https://doi.org/10.1002/(SICI)1520-6656(199924)13:1 3.0.CO;2-L
• "Improving Customer Service and Profitability Through Customer Intervention in Service Relationships," Journal of Relationship Marketing. https://doi.org/10.1080/15332660802508505
• 37signals, "Kill overkill." https://basecamp.com/overkill

This is an educational decision framework, not individualized financial or legal advice. Contractual termination, regulated industries and employment-related customer arrangements may require professional advice.

This article should be read alongside the nearest SOE Knowledge Objects covering the same problem, stage, and next decision. Use those related pages to compare evidence, context, and implementation choices rather than treating this page as a standalone rule.

Related reading: \2 · \2