Every buyer knows to check customer concentration. You pull the revenue by customer, look for any one name above a threshold, and relax when the list is spread out. This lesson is about the concentration that test cannot see: revenue that is spread across many customers but held by very few people. When those people leave, the customers go with them, and the income statement you underwrote turns out to have been a forecast about two employees' loyalty.
The case is Alan Brewer's, from the Buyer Beware failure record. His sellers were honest, his profits were real, his price was fair, and his diligence ran a hundred days. He lost the company anyway.
Principle
Key-person risk is about who owns the customer relationship, not what the org chart says. The org chart tells you who reports to whom. The relationship map tells you who the customer calls. Only the second one predicts where the revenue goes when someone quits.
Put it as a rule you can run: for every material account, name the human the client actually calls, then ask what binds that human to the company. If the honest answer is goodwill, you are not buying that revenue. You are renting it, on terms the employee can end with one day's notice.
Customer concentration is a recognized cause of failure even in the better-resourced world of traditional, investor-backed search funds. A 2026 IESE note built from investor observations lists customer concentration among the leading causes of search fund failure, with risk assessment in diligence as the first remedy (L15). That note describes funded searches, not self-funded deals, but the mechanism is the same and your margin for error is smaller. Your version of the test has to count concentration in people as well as in customers.
Case
Alan Brewer spent his career in finance at large firms. He was, in his own telling, the careful one in every room. He spent about eight months testing the path before he quit his job in 2019, talking to thirty-five or forty people who had already bought companies. He searched only near Denver and deliberately chose a white-collar business he believed he could manage.
In January 2020 a broker brought him a company that built custom training programs for large corporations. Fortune 500 names sat on the client list. The contracts were master service agreements, standing deals that make each new project easy to start. More than 90 percent of revenue came from repeat clients. The two owners stayed behind the scenes and did almost no client work.
The price was $3 million, about 4.2 times yearly profit. A government-backed loan funded most of it, and Alan signed a personal guarantee. About a third of the price sat in a seller note that would shrink if sales fell after closing. Diligence ran about a hundred days. Outside accountants confirmed the profits were real; they were. A labor lawyer reviewed hiring risk. Alan stress-tested a 50 percent revenue decline and concluded he would survive it.
One term should have stopped him. He was not allowed to talk to employees until a week before closing, and in that week he met exactly two: the senior salespeople, Gary and Renee. Between them they touched nearly every client. Gary's relationships went back as far as forty years. The client list was spread out. The relationships were not. He was warned only that Gary "could be difficult at times."
He closed in September 2020. Within a month Gary was openly hostile to the new owner and the former owners, while remaining charming with clients. Alan was afraid to act, because firing Gary meant the clients might follow him out. Year one revenue grew more than 20 percent, but profit sagged and change stalled. About eighteen months in, after two employees came to him in tears about how Gary treated them, Alan fired him on his labor lawyer's advice.
Within about two months Gary had a competing firm and was calling clients he had served for decades. In Colorado, a non-compete against an ordinary employee is close to unenforceable, and nothing in the company's paper could stop him. Clients did not announce they were leaving; they went quiet. In a project business, silence looks like a normal lull. Revenue for 2022 was $3.6 million. In October 2023 Renee quit with one day's notice and took most of her clients with her. Revenue fell to $1.4 million for 2023 and to about $1 million in 2024. Alan's own split of the blame: roughly 60 percent the poaching, 40 percent a soft economy. He also concluded, too late, that his niche had almost no real salespeople and grew mostly by luck, so there was little new business to win back.
The company stopped operating in March 2025. Payroll cleared first. About $1.5 million was still owed on the government-backed loan, and his signature sat under all of it. When last interviewed for the book, he was preparing to file for bankruptcy himself, because the guarantee had made the company's debt his own. Lesson 8.4.3 covers the bankruptcy chapters a failed business and its guarantor face.
The seller note worked exactly as written: sales fell and the note was cut by more than half. It saved no one. The sellers could not help win the clients back, because they had never known them. The men who sold the company had never owned its revenue either.
Three points to carry forward:
- Nobody lied. A quality of earnings report would not have caught this, and under today's SBA rules one is required when the purchase price is $3 million or more. The numbers were true. The risk lived in who held them.
- The stress test was sized to the wrong disaster. A 50 percent haircut models a bad market. It does not model the departure of the revenue's owners. The real decline was closer to 75 percent.
- The access restriction was the tell. One week of employee access on a $3 million purchase meant the most important diligence was scheduled after the decision was effectively made.
Steps
Step 1: Build the relationship map
Take the accounts that make up roughly 80 percent of revenue. For each one, record:
| Field | What you are asking |
|---|---|
| Account and annual revenue | From the revenue-by-customer report, tied to invoices |
| Who the client calls | A named person at the company, not a department |
| Who at the client makes the call | The buyer, and whether that person has changed recently |
| Relationship age | How long that pair has worked together |
| Backup contact | A second person at the company the client actually knows |
| What binds the employee | Equity, deferred pay, retention bonus, non-solicitation agreement, or nothing |
The Yale case on customer attrition makes a related point: analyzing which customers leave, and when, explains what really sits under revenue performance (M62). The relationship map is that analysis pointed forward. You are asking which customers would leave if one person did.
Step 2: Measure concentration in people
Add up revenue by relationship holder. Then treat each person exactly as you would treat a customer of the same size. If one salesperson touches 40 percent of revenue, you have the same exposure as a 40 percent customer, with one difference: a customer usually gives notice. The thresholds in the Walk-Away Doctrine (one customer over 40 percent at the screen; over 30 percent with no contract at the conviction gate) are a sensible starting point for people too.
Include the owner. In many small businesses the largest relationship holder is the seller. That is red flag RF-04 (owner dependency). When it is an employee, it is RF-07 (key-employee risk). When it is a client that pays through one person, it is also RF-01 (customer concentration), counted in people.
Step 3: Ask what binds each person
For every relationship holder above your threshold, write down what keeps them after closing. Then test each binding:
- Non-compete. Enforceability is a matter of state law and it varies widely. Research on founder replacement used changes in non-compete law across 14 U.S. states between 1995 and 2016 as a natural experiment (E8), which tells you how often the ground moves. Ask a local employment attorney what is enforceable in your state before you count on any covenant.
- Non-solicitation. A promise not to solicit clients or staff is often narrower and sometimes more enforceable than a non-compete. Have counsel confirm.
- Money that arrives only if they stay. A retention bonus paid in installments over the first one to two years, or a share of growth on accounts they keep, is a binding you control.
- Goodwill. Goodwill is not collateral.
Step 4: Get access before you commit
Negotiate structured conversations with the revenue-critical people before your letter of intent goes exclusive, or as an early condition inside it. A seller's confidentiality concerns are real; meet them with a script, a non-disclosure agreement, and the seller in the room if need be. If the seller will not let you meet the people who hold the revenue, that is a price factor, not a custom. Pay less for what you cannot see, or walk.
When you do meet them, ask the employee directly: which clients would follow you if you left, and what would make you stay?
Step 5: Size the downside from the map, not a round number
Run your debt service test with the top relationship holders gone and their accounts gone with them. Not a 10 or 50 percent haircut: the actual accounts on the map. If coverage fails, either the structure changes or the deal does.
Illustrative arithmetic: a business earns $600,000 a year before debt service, and two salespeople hold accounts producing $350,000 of that contribution. Debt service is $400,000. With both gone, earnings fall to $250,000 against $400,000 of payments. The business cannot carry its loan, and the personal guarantee means you carry it instead.
Step 6: Structure around the people you cannot replace
Tools that help, all to be drafted by your attorney:
- Retention agreements for named employees, signed at or before closing and funded from the sources and uses.
- A seller who shares the loss. Under SBA SOP 50 10 8.1, effective October 1, 2026 (Y25), seller earnouts are prohibited in 7(a) acquisitions. The one contingent tool the SOP names is a buyer rebate from the seller based on business performance, and any rebate must be applied to pay down the principal of the 7(a) loan. The SOP does not name a seller note whose balance moves with retention of named accounts, as Alan's did; a lender decides whether to allow one. Remember Alan's lesson: a note that shares the loss does not prevent it.
- A seller transition. Under SOP 50 10 8.1, when you buy 100 percent (an Initial Acquisition) the seller may stay only as a paid consultant, for up to 24 months. That helps only if the seller actually holds relationships. Ask.
Step 7: Re-run the map the week before closing
Get the roster and revenue by salesperson again just before the wire. People leave during diligence, and the company you priced at the letter of intent is not always the company you receive. Make any departure of a named relationship holder a disclosure the seller must make, with a price consequence written into the purchase agreement.
Step 8: Install an early-warning system on day one
In project businesses, a dying account sounds like a quiet month. From the first week, assign every named account an owner-level contact cadence and a "last touch" date. Introduce yourself to every key client contact in the first sixty days, ideally alongside the employee who holds the relationship.
Apply it to Coastal Comfort HVAC
In the fictional Coastal Comfort case, the owner, Ray Delgado, runs sales, dispatch and the largest jobs himself. He has four salespeople and estimators and no second-in-command. Map it: which commercial customers and builders call Ray, which call an estimator, and who else at the company each one knows by name. In this case the map finds two holders. Ray personally holds the relationships on the six largest commercial accounts. The homebuilder behind the FY2025 townhome project turns out to rest on one estimator's friendship, with no master agreement (lesson 7.3.1). Revenue held only by Ray leaves when his consulting period ends unless you build the second relationship during it. Revenue held by the estimator leaves whenever the estimator does.
Deliverable
A one-page Relationship Map for the deal:
- Every account in the top 80 percent of revenue, with the named relationship holder, the client-side contact, relationship age, backup contact, and binding.
- Revenue concentration by person, with any holder above your threshold flagged as RF-04 or RF-07.
- The downside case: debt service coverage with the flagged holders and their accounts removed.
- Your structural response for each flagged person: retention agreement, seller note or rebate terms your lender has approved, consulting plan, or walk.
- The answer, in one sentence, to the question Alan Brewer would ask: if the key person quits tomorrow with one day's notice, where does the revenue go?
This lesson is education, not legal advice. Non-compete, non-solicitation and retention terms depend on state law; have an employment attorney licensed in your state review them before you rely on any of them.