LISTEN
Listen to this perspective
Written by Oleg Cohen. Companion narration uses the Brian voice.
The company has attractive earnings, loyal customers, an experienced team and a credible growth story. The financial model supports the investment. The value-creation plan names sensible opportunities: improve pricing, expand geographically, integrate acquisitions, strengthen the commercial organization.
Then someone asks how an unusual quote gets approved.
The owner handles it. Who knows which customer commitments can’t be disturbed? The owner and the head of operations. Who understands why one apparently low-margin account is worth keeping? The controller, if you ask the right way. Who knows which supplier to call when the formal purchasing process fails? The same operations leader.
The business has systems and processes. It may even appear to have a deep management team. But important parts of how it actually operates live in the judgment, memory, relationships and constant intervention of a handful of people.
That isn’t evidence of a bad business. It may be evidence of an exceptionally resourceful founder and core team. It does raise a different investment question.
Are we underwriting capabilities the enterprise has, or the continued performance of a few exceptional people?
Key-person risk is only part of the problem
Private equity already watches for founder dependence, key-person risk and management depth. Alok Gupta of Loki Group describes it well. In owner-operated businesses, he writes, the founder is often the operating system of the company, and standard diligence, built to evaluate documented information, rarely surfaces it. His test is simple: if the seller took a three-month sabbatical starting tomorrow, what would break?
That test asks what happens if a critical person is absent. Absence is only one way the dependency bites.
What happens if everyone stays, and the sponsor asks the same small team to run the business, produce more reporting, improve margins, open locations, integrate acquisitions and build the next management layer? What happens if the founder is willing to delegate, but nobody else has the context to use that authority well? What happens when a new executive understands the investment thesis but not the operating judgments that sustain today’s earnings?
Gupta calls the narrower version operational concentration risk. I use capability concentration risk for the broader exposure: the extent to which value-critical knowledge, judgment, relationships, authority and coordination depend on a few people, and not on capabilities the enterprise can repeat. The question isn’t how many people are on the organization chart. It’s how many can reliably produce the outcomes the business needs, under the conditions it actually meets.
Earnings and execution rely on different capabilities
A business can earn good money through arrangements that are hard to reproduce. An experienced founder spots a bad deal before anyone else, rescues an account through a personal relationship, or reconciles conflicting priorities with years of context. Those earnings are real. The question is whether they survive a change in ownership, leadership, volume or complexity.
Consider a hypothetical regional service company. The owner handles unusual pricing, the operations leader knows how to recover troubled jobs, and the controller knows which billing anomalies matter. Customers get good service because these three coordinate constantly, mostly outside formal workflows.
The sponsor sees a chance to expand into two territories and add an acquisition. The opportunities may be commercially sound. But the plan quietly assumes capabilities the business has never demonstrated: launching branches, developing managers, transferring customer context and coordinating exceptions across a larger organization. The people who make today’s business work are also expected to build its next operating model.
That is an underwriting gap: the distance between the capability the thesis assumes and the capability the organization can deliver without concentrated personal intervention. McKinsey makes a related point about operational diligence. Sizing the improvement opportunities isn’t enough; investors have to understand the operations and whether the leadership needed to execute the plan is in place. I’d go one step further and examine the system around those leaders as carefully as their résumés. Management attention has a capacity limit, and underwriting should make it visible.
The executive inherits authority before understanding
The problem becomes urgent when a sponsor appoints a new CEO, president, COO or CFO, and that happens often. Bain’s Hugh MacArthur notes that new owners replace somewhere between half and 70 percent of portfolio company CEOs during the hold period.
The incoming executive gets the financials, the thesis, the organization chart, system access, board expectations and a transition schedule. They are accountable almost immediately. But authority transfers faster than understanding. The new leader may not know why a customer gets an accommodation, why a purchasing rule is sometimes bypassed, which employee carries responsibilities that appear in no job description, or why an obvious initiative failed three years ago.
A newly assembled platform makes this harder. There may be no single operating history to inherit. Acquired businesses can use the same term to mean different things and rely on different informal arrangements. Standardizing their reports doesn’t resolve those differences.
I call this the operating inheritance gap: the distance between taking responsibility for a company and understanding it well enough to run and change it. A talented executive can make reasonable decisions from an incomplete model of the business. A founder’s apparent pricing inconsistency may hold valuable customer history, or it may simply be poor discipline. An informal scheduling habit may protect an important service commitment, or hide avoidable waste. The new leader needs a way to tell the difference before preserving everything or replacing everything.
Hiring the executive doesn’t, by itself, transfer the company.
Knowledge transfer has to happen inside real work
Interviews, handover documents, process maps and recordings help. They rarely capture what matters most. Experienced operators can’t recite their judgment as a set of rules. A situation triggers a memory; a detail changes its meaning.
Suppose a service manager wants to offer a credit after a repeated failure. Policy sets a limit. The founder nevertheless asks about the customer’s equipment, the technician assigned, a previous commitment and a repair that was never finished. The right response depends on how those facts relate, not on the credit amount.
The useful record is the situation, the evidence noticed, the interpretation, the alternatives considered, the authority used, the action and what happened afterward. That’s worth far more than “use judgment with important customers.” It helps the next manager recognize a similar situation, spot missing evidence and know when to escalate. As I argued in Talk to Susan, the aim is to make what the best people know usable by others.
It shouldn’t harden into an unquestionable rule. The founder’s interpretation may be incomplete or out of date, so good capture includes disagreement and validation, not just transcription. And no record transfers personal trust. Some transitions still need joint customer visits, introductions and time.
Company meaning is what makes AI useful here
This is where modern AI can help, but only if it is grounded in the particular company. A general-purpose model can summarize a transition interview. It can’t safely assume what the company’s words mean, which source is authoritative, who can act, or how a decision affects commitments and economics.
Take a dealership. “Ready” might mean mechanically complete, photographed, title-cleared or available for retail sale. A model that treats those as the same can confidently recommend promoting a car that can’t yet be delivered. In recurring field service, an “active account” may be billable, scheduled, paused or awaiting reinstatement. Count the records without those distinctions and you misstate demand and capacity.
A company-specific model of the business connects language to operating reality: customers, assets, services, commitments, people, locations, work, exceptions, decisions, authority, costs and outcomes. It has to be built from the company’s actual records, documents, contracts, interviews and observed decisions, with conflicts and unknowns left visible. The practical test is whether those distinctions change the answer when they should.
With that foundation, an incoming executive can ask better questions:
- Which capabilities in our growth plan still depend on the founder?
- Which operating practices do we understand well enough to change?
- Where does the formal process differ from the actual work, and why?
- Which decisions can managers take now, and with what evidence?
- What have we learned since diligence that changes the thesis?
The value isn’t that AI remembers more documents. It’s that people can reason and act with more of the company’s context in front of them.
From concentrated judgment to governed capability
Capturing knowledge is a start. Transfer has happened only when someone else can use it well in practice.
Return to the service company. Instead of routing every unusual credit to the owner, it builds a customer-recovery capability. The service manager sees the relevant commitments, visit history, prior exceptions, costs and escalation criteria. The company defines what the manager may approve, when the owner must be involved and what evidence is required. An AI-supported workflow assembles the history, flags unresolved commitments, suggests alternatives and prepares a proposed action. Within explicitly authorized limits, an agent might open an internal task or request missing evidence. Consequential customer, financial or contractual actions still go through the authority and approvals the business sets.
Then the outcome feeds back. Did the action resolve the problem? Did it create rework? Was the customer retained, and was the decision economically sound? Should the authority limit or the guidance change?
The capability is larger than the agent. It includes people, judgment, evidence, authority, work, systems, economics and learning. The aim isn’t a digital imitation of the founder. It’s an organization that can apply more of what it knows without the same few people personally carrying every decision.
What investors and incoming executives should test
A useful assessment ties each dependency to an investment assumption or operating outcome. A long list of undocumented practices isn’t enough.
Start with the capabilities that protect earnings or enable the value-creation plan. For each, ask who supplies the judgment, what information they use, what authority they exercise, who could substitute, and what happens as demand or complexity grows. Then separate presumed transferability from demonstrated transferability. Can another manager handle a representative exception with the right evidence and authority? Can a new location reproduce the outcome? Does the process continue when the founder is unavailable?
Measure dependency and performance together: escalation frequency, decision delays, rework, avoidable errors, time to proficiency, customer continuity and the economics of the capability itself. A falling escalation count isn’t progress if people have simply stopped raising problems. And better margins after a handover don’t prove the transfer caused them. Baselines and comparable cases make the evidence credible.
For an executive’s first 100 days, that suggests a sequence: understand the operating dependencies, validate the company model, transfer selected decisions under controlled conditions, and check whether outcomes hold. Urgent safety, liquidity or compliance problems may need action sooner. The goal isn’t to map everything before doing anything. It’s to understand enough to improve something consequential without discarding the context that makes it work.
Preserve the advantage, reduce the dependency
Not all concentration is bad. Some businesses legitimately depend on scarce expertise, credentials, reputation or unusual relationships, and in a small company full redundancy may not pay. The question is whether the concentration is understood, proportionate and consistent with the plan.
An earnout, a retention agreement, an advisory period or a new hire can address part of the exposure. But keeping someone available isn’t the same as enabling the organization to do their work. Shifting the dependency from the founder to the newly hired CEO isn’t institutionalization either.
The founder should be a partner in the transition. Explain what is being captured, who can see it and how it will be used. Protect employee privacy, customer confidentiality and contractual restrictions. Sponsor oversight doesn’t imply access to every conversation.
Enterprise value needs evidence of enterprise capability
At exit, growth alone doesn’t answer the transferability question. A larger company may still depend on the same small team to hold it together.
A stronger account shows what changed: a capability that once depended on the owner now runs across defined roles, with supporting knowledge, explicit authority, tested execution, measured outcomes and learning that survives leadership changes. That doesn’t guarantee a higher multiple, which also depends on the market, the business and the deal. It’s a more substantive answer than saying the company professionalized or adopted AI. For private credit the same reality matters through a different lens: continuity, cash generation and the capacity to respond when performance slips.
People will always matter. The question is whether their knowledge and judgment are becoming an asset the enterprise can use, develop and carry forward, or whether the enterprise must keep depending on them to produce the same result.
Private capital should know whether it is underwriting an enterprise or the continued heroics of a few exceptional people. And the executive hired to run it should inherit more than the responsibility.
Where I sit: I advise through Kainora and am building Lattice, a platform aimed at this problem, preserving a company’s context and learning so its people can act on it. It’s in production with one company and in pilots with several more. Whether it reduces dependency is something to prove in bounded operating tests, not to assume.
What have you seen fail, or succeed, when a founder’s judgment had to become something the rest of the company could use?
Sources and further reading
- Alok Gupta, “The Seller Who Was Also the ERP,” Loki Group, February 1, 2026
- Jose Luis Blanco, William Bundy, Matthew Maloney and Jason Phillips, “Bridging private equity’s value creation gap,” McKinsey, April 12, 2024
- Hugh MacArthur, “A Private Equity Fund Just Bought Your Company … Now What?,” Bain & Company, November 21, 2025