Knowledge Due Diligence

Financial and legal diligence can confirm what a business owns, owes and earns. They do not reveal where its judgment lives, how fragile that knowledge is, or whether the buyer can reproduce it.

At closing, an acquirer receives shares, contracts, bank accounts, intellectual property, equipment, source code, customer records and access credentials.

It does not automatically receive the company’s ability to make good decisions.

Consider a hypothetical specialty maintenance business generating $40 million in revenue and $6 million in EBITDA.

The financial statements are clean. Customer concentration is acceptable. Contracts have been reviewed. The equipment is in good condition. The CRM contains years of account history. Operating procedures are stored in a shared drive. The management team appears stable.

The acquisition closes.

Over the following months, several things begin to deteriorate.

Estimates become less accurate because the senior estimator had been adjusting quotes based on patterns that were never written down. Crew utilization falls because the operations manager knew which technicians could work together, which sites created hidden delays and which customers would tolerate a schedule change. A major account leaves after a routine service failure because the founder had always known when to call the customer personally before a small issue became a commercial problem.

None of the information presented during diligence was necessarily false.

The buyer correctly verified the assets, liabilities, contracts and historical cash flows.

What it failed to identify was the knowledge system that produced them.

The balance sheet tells a buyer what the company owns.

The organizational chart tells a buyer who works there.

Neither explains where the company knows what it knows.

Every acquisition therefore involves the purchase of two companies.

The first is the legal and economic company. It consists of the entities, contracts, assets, liabilities and rights that can be transferred at closing.

The second is the cognitive company. It consists of the accumulated judgment, memory, relationships, routines and decision-making patterns that allow the business to function.

The legal company transfers with signatures.

The cognitive company does not.

It must be understood, retained, transferred or rebuilt.

Most acquisition processes examine the first company in extraordinary detail and treat the second as an integration problem to be addressed after the transaction.

That is too late.

Knowledge should be diligenced before the price is finalized.

The part of the business that cannot be uploaded to the data room

Some organizational knowledge is explicit.

It can be written, stored and reviewed:

• contracts
• operating manuals
• pricing tables
• software documentation
• customer records
• financial models
• policies
• process diagrams
• product specifications

Other knowledge is tacit.

It appears in judgment, pattern recognition, relationships and experience. It determines how people respond when the standard process is incomplete, the available information is ambiguous or the situation does not match the manual.

Organizational research has long distinguished between explicit knowledge and knowledge that is difficult to articulate. Nonaka described organizational knowledge creation as a continuing interaction between tacit and explicit knowledge. Later work on knowledge transfer similarly emphasized that tacit knowledge requires different transfer mechanisms from information that can simply be written down and distributed.

Tacit knowledge is not mystical.

Much of it is simply unobserved decision logic.

It is the experienced salesperson recognizing that a prospect asking a seemingly technical question is actually concerned about implementation risk.

It is the plant manager hearing a small change in a machine and knowing that production should stop before a dashboard reports a failure.

It is the account manager understanding that one customer’s complaint can be resolved with a credit, while another customer interprets the same response as evidence that the supplier does not understand the relationship.

It is the engineer knowing which apparently harmless part of a legacy system should not be modified before the end of the quarter.

It is the founder recognizing that the company should reject a large contract because the customer’s operating behavior will make the account unprofitable.

This knowledge often has no formal owner.

It may be distributed across conversations, habits, inboxes, personal notes, meeting history and the memories of people who have been solving similar problems for years.

The company benefits from it every day, but may never have identified it as an asset.

The acquisition exposes that weakness.

Existing diligence does not answer the right question

A sophisticated transaction can involve financial, legal, tax, commercial, operational, technology, cybersecurity, regulatory and human-capital diligence.

Each workstream answers important questions.

Financial diligence asks whether the earnings are real.

Legal diligence asks whether the company owns what it claims to own and whether hidden obligations exist.

Commercial diligence asks whether customers will continue buying and whether the market supports the plan.

Technology diligence asks whether the systems are secure, scalable and maintainable.

Human-capital diligence commonly examines organizational structure, compensation, employment obligations, leadership capacity, retention risk and workforce design.

None of these disciplines, by itself, fully answers a more fundamental question:

What must this organization continue to know in order to reproduce its performance after control changes hands?

Human-capital diligence may identify the people a buyer wants to retain.

Knowledge due diligence asks what would leave with those people.

Operational diligence may document the process.

Knowledge due diligence asks which exceptions make the process work.

Technology diligence may review the architecture and codebase.

Knowledge due diligence asks who understands why the architecture evolved that way, which constraints are still real and which apparently irrational decisions protect the business from an old failure.

Commercial diligence may confirm that a customer is satisfied.

Knowledge due diligence asks who understands the history of that relationship, what promises were made outside the contract and which future action could unexpectedly destroy trust.

The difference is subtle but material.

Traditional diligence often evaluates the visible structure of the company.

Knowledge due diligence evaluates the organization’s ability to understand and act.

The concept itself is not entirely new. Knowledge-management researchers have previously argued for examining intellectual assets and organizational knowledge during transactions. It has not, however, become a standard, decision-grade workstream with a consistent link to valuation, deal structure and integration planning.

That is the gap.

What knowledge due diligence should mean

Knowledge due diligence is the systematic assessment of the knowledge required to reproduce a company’s operating performance after ownership changes.

It should determine:

• What knowledge is critical to value creation.
• Where that knowledge currently resides.
• How concentrated and fragile it is.
• Whether it can be observed, transferred or reproduced.
• How long the transfer will take.
• What happens if the current knowledge holders leave.
• How the findings should affect price, terms and integration.

The objective is not to document every activity in the company.

That would produce an enormous archive with little decision value.

The objective is to locate the knowledge that supports the acquisition thesis.

If the buyer’s case depends on customer retention, the diligence should identify the knowledge that protects those relationships.

If the case depends on margin expansion, it should identify the operating judgment currently preventing waste, rework, discounting or poor resource allocation.

If the case depends on product development, it should identify the technical and customer knowledge required to make good roadmap decisions.

If the case depends on geographic expansion, it should identify which capabilities are genuinely repeatable and which rely on local relationships or individual experience.

Knowledge diligence should begin with the value-creation plan, not with the employee directory.

The six forms of knowledge a buyer is actually acquiring

The knowledge inside a company can be divided into six practical categories.

Knowledge type	What it contains	Typical risk
Procedural knowledge	How work is actually performed, including workarounds and exceptions	The documented process does not reproduce the real process
Judgment knowledge	How experienced people make decisions under uncertainty	Replacements follow the rules but make consistently worse decisions
Relational knowledge	Trust, history and context within customer, supplier and regulatory relationships	Contracts survive while the commercial relationship deteriorates
Causal knowledge	Understanding of what truly drives outcomes and why	The buyer changes a variable without understanding second-order effects
Historical knowledge	Prior experiments, failures and reasons behind current choices	The new owner repeats old mistakes or removes necessary constraints
Coordination knowledge	Informal networks, escalation paths and sequencing of work	The formal organization remains intact while execution slows down

These forms of knowledge behave differently.

Procedural knowledge can often be captured through observation, process reconstruction and scenario testing.

Judgment knowledge is harder. It may require reviewing a series of past decisions, identifying the signals considered by the expert and comparing those decisions with outcomes.

Relational knowledge cannot be reduced to a contact list. It may require joint customer meetings, explicit transfer of trust and a period during which the former relationship owner remains visibly involved.

Historical knowledge requires more than asking, “Why do we do it this way?”

People often remember the current rule but not the event that created it. A useful investigation must trace the sequence of decisions, failures and constraints behind the present operating model.

Coordination knowledge is frequently hidden in informal behavior. The organization chart may show reporting lines, while the real company operates through a different network of people who know whom to involve when something goes wrong.

A buyer that does not distinguish these categories will tend to prescribe the same solution for every problem: document the process and retain the senior employees.

That is insufficient.

Key-person risk is only a crude proxy

Deal teams already understand key-person risk.

They ask whether revenue depends on the founder, whether an important engineer might leave or whether a major customer relationship belongs to one salesperson.

This is useful, but incomplete.

Key-person analysis identifies a person.

Knowledge analysis identifies the capability that depends on that person.

The distinction matters because a single individual may hold several different forms of knowledge, each requiring a different response.

A founder might hold:

• the trust of three major customers
• the logic behind the company’s pricing model
• the history of failed product expansions
• the criteria used to evaluate senior hires
• the informal authority to resolve conflicts between departments

A two-year retention agreement does not automatically transfer any of these capabilities.

It merely keeps the person available for a period of time.

Retention buys time.

It does not create transfer.

The same is true of an earnout. An earnout may align the seller economically with short-term performance, but it does not guarantee that the buyer understands how the seller is producing that performance.

A company can meet its earnout targets while remaining just as dependent on the seller as it was on the day of closing.

Knowledge due diligence should therefore replace the broad idea of key-person risk with a more precise assessment of knowledge at risk.

Knowledge at risk can be thought of as the exposure created when a value-critical capability depends on knowledge that is concentrated, opaque and difficult to transfer within the transaction horizon.

It can be assessed using four dimensions:

Knowledge at Risk = Value Criticality × Concentration × Transfer Friction × Time Sensitivity

This is not an accounting formula.

It is a decision framework.

Value criticality asks how much revenue, margin, customer retention or operational continuity depends on the capability.

Concentration asks how many independent people or systems can currently perform it.

Transfer friction asks how difficult it would be for another person or organization to learn it.

Time sensitivity asks how quickly value begins to deteriorate if the knowledge becomes unavailable.

Consider a hypothetical business with $6 million of EBITDA acquired at an eight-times multiple.

Suppose $1.5 million of that EBITDA depends on pricing and scheduling decisions made by two senior operators. Their work is poorly documented, and no one else can consistently reproduce their decisions.

If those capabilities disappear, the immediate earnings exposure is $1.5 million. At the acquisition multiple, the implied value exposure is $12 million.

In that context, a substantial retention package, a six-month parallel operating period and a structured knowledge-transfer program are not administrative expenses.

They are value protection.

The accounting treatment itself illustrates the problem. Under IFRS 3, identifiable acquired intangible assets are recognized separately from goodwill, but an assembled workforce is not treated as a separately identifiable asset. Its value is generally subsumed into goodwill. Internally generated goodwill is also not recognized as an asset under IAS 38.

The economics of organizational knowledge therefore matter even when the balance sheet cannot isolate them cleanly.

A practical knowledge-diligence process

A useful knowledge-diligence process should not begin by asking employees to document everything they do.

It should begin by identifying the decisions that make the acquisition valuable.

1. Start with the deal thesis

The buyer should state clearly what must remain true after closing.

For example:

• customers must renew at the historical rate
• gross margins must remain stable
• project delivery must not slow down
• the product roadmap must continue
• a founder relationship must become an institutional relationship
• a successful local operation must be replicated nationally
• automation must reduce cost without damaging service quality

These assumptions become the starting point for knowledge mapping.

2. Identify value-critical decisions

Every operating model contains a relatively small number of recurring decisions with disproportionate economic impact.

Which opportunities should sales reject?

When should a customer receive an exception?

How should a complex job be priced?

Which quality issue requires production to stop?

When should an account be escalated?

Which product request indicates a wider market need?

Which apparent cost saving would create a larger downstream expense?

The diligence team should identify these decisions and estimate their relationship to revenue, margin, retention, risk and working capital.

3. Trace how those decisions are actually made

The investigator should not rely solely on process descriptions.

People tend to explain how work is supposed to happen.

The relevant evidence is how work actually happened.

That means reviewing real examples:

• pricing changes
• customer escalations
• support cases
• project overruns
• lost deals
• quality failures
• supplier disputes
• incident reports
• product decisions
• internal approvals
• postmortems
• exceptions to standard policy

The most revealing question is often not, “What is the process?”

It is:

Tell me about the last time the process was not enough.

4. Map the knowledge holders

For each critical capability, the buyer should know:

• who currently performs it
• who understands why it is performed that way
• who can act as a substitute
• what evidence or documentation exists
• how often the capability is used
• how long it would take another person to become competent
• what would happen during the learning period

The formal owner may not be the real knowledge holder.

A vice president may approve a decision that is effectively designed by a long-tenured analyst.

A founder may appear central to a customer relationship, while the customer actually trusts an implementation manager.

A chief technology officer may own the architecture, while one senior engineer understands the fragile dependency that could create an outage.

Knowledge diligence must follow the decision, not the title.

5. Run the absence test

For every critical knowledge holder, ask:

What happens if this person becomes unavailable tomorrow?

Then examine the consequences after 30, 90 and 180 days.

Would revenue decline?

Would decisions slow down?

Would quality deteriorate?

Would customers notice?

Would the company remain compliant?

Would the team continue operating but accumulate invisible errors?

The last possibility is particularly dangerous.

Some knowledge failures do not produce an immediate crisis. They produce a gradual decline in decision quality that only appears later in margins, churn, rework or lost opportunities.

6. Test reproduction, not documentation

A document should not be considered evidence of successful transfer.

The test is whether another person can reproduce the result.

Give a successor a realistic customer escalation and observe the response.

Ask another estimator to price a complex project and compare the assumptions.

Have a second engineer explain the failure modes of a critical system.

Ask a new account owner to conduct a customer review without intervention from the previous owner.

Run operational simulations.

A company has not transferred knowledge because a manual exists.

It has transferred knowledge when another person can make a sufficiently good decision.

7. Convert the findings into transaction decisions

The output should not be a general report about culture or employee sentiment.

It should be a decision document linking knowledge risk to:

• valuation
• retention
• earnout design
• transition periods
• integration sequencing
• management continuity
• operating budget
• customer handovers
• training requirements
• systems migration
• the first 100-day plan

The purpose of diligence is not simply to know more.

It is to make a better transaction decision.

AI changes what is economically feasible

Organizational knowledge has always mattered.

What is changing is the cost of investigating, capturing and operationalizing it.

Historically, a serious knowledge-transfer exercise required extensive interviews, observation, transcription, process mapping and manual synthesis. This made it expensive, slow and difficult to scale across an entire company.

Artificial intelligence can reduce that friction.

A properly governed system can analyze large bodies of operational material such as meeting transcripts, support histories, project records, customer interactions, technical discussions and decision logs.

It can identify recurring exceptions.

It can compare formal procedures with actual behavior.

It can detect where different employees describe the same process differently.

It can generate targeted follow-up questions.

It can organize examples by capability, customer, decision type and outcome.

It can help create decision trees, case libraries and searchable institutional memory.

Most importantly, it can examine what experts do, not only what they say they do.

A 2026 preprint studying expert math tutors found that knowledge externalized by language models from expert conversations improved downstream decision quality and helped novices approach expert performance. In that specific experimental setting, the extracted knowledge outperformed knowledge that experts had articulated directly. The result is promising, but it remains a domain-specific preprint rather than proof that tacit knowledge can be reliably extracted across businesses.

A separate 2025 study explored an AI agent designed to navigate simulated organizational structures and gather fragmented knowledge through repeated conversations. It demonstrated the potential of agent-based knowledge elicitation, but the environment was simulated and should not be confused with validation inside a real acquisition.

These limitations matter.

An AI system cannot download a person’s experience.

It can misunderstand context, infer false patterns and create a polished but incorrect explanation. It may hide disagreement by synthesizing several incompatible views into a single answer. It may also expose confidential, personal or privileged information if the system is poorly governed.

Every extracted claim should remain connected to its source.

Every important operating conclusion should be validated by people with direct experience.

Conflicting explanations should be preserved, not automatically resolved.

The strongest near-term use of AI is not to create a synthetic replacement for an employee after that employee leaves.

It is to improve the quality, reach and speed of the transfer while the employee is still present.

AI can turn knowledge capture from a one-time exit interview into a structured operating process.

Knowledge diligence should change the price and the terms

Two companies can produce the same EBITDA and still represent very different acquisition risks.

In the first company, critical decisions are distributed across a capable management team. Customer history is maintained in usable systems. Operational exceptions are reviewed. Important technical decisions are recorded. Multiple people can perform each critical function. Departures are inconvenient but manageable.

In the second company, pricing exists in the founder’s head. Customer context lives in private inboxes. One analyst maintains the reporting model. One engineer understands the production system. Operating procedures describe the normal case but not the exceptions. Employees know the company works, but cannot fully explain why.

The financial statements may look similar.

The quality of the earnings is not.

The first business has greater knowledge resilience.

The second contains a hidden concentration risk.

A buyer should be prepared to pay more, or accept less restrictive terms, for a company whose critical knowledge is distributed, observable and reproducible.

A knowledge-fragile business may require:

• a valuation adjustment
• a longer transition period
• targeted retention arrangements
• a seller earnout
• greater integration reserves
• delayed systems consolidation
• customer-specific handovers
• parallel operations
• pre-closing transfer milestones

The structure should follow the knowledge risk.

Retention packages should be based on the knowledge map, not only on hierarchy.

A lower-level employee who controls a critical operational capability may be economically more important to the transition than a senior executive with a visible title.

Earnouts should also consider transfer outcomes.

A seller should not receive full credit merely because the company meets short-term earnings targets while remaining entirely dependent on the seller’s personal involvement.

Useful transfer milestones could include:

• successful handover of major customer relationships
• independent execution of critical operating scenarios
• completion of a validated decision library
• demonstrated redundancy for key functions
• delivery of historical context behind material systems or processes
• replacement managers achieving defined performance thresholds

The purpose is not to make transactions unnecessarily complicated.

It is to align consideration with the transfer of the asset the buyer believes it is purchasing.

The seller should begin years before a transaction

Knowledge diligence is not only a buyer’s tool.

It should become part of exit preparation.

Founders frequently claim that a business can operate without them because they can take a vacation or avoid day-to-day involvement.

That is not a sufficient test.

The company may continue operating because employees defer major decisions, contact the founder privately or avoid taking actions that would expose the dependency.

A truly transferable business can do more than continue routine activity.

It can respond to exceptions, protect relationships, allocate resources and make consequential decisions without relying on one person’s undocumented judgment.

A seller preparing for a future transaction should therefore invest in knowledge resilience.

That means capturing decisions, not only procedures.

A useful operating record should explain:

• what was decided
• what information was available
• which assumptions mattered
• what alternatives were rejected
• why an exception was made
• what outcome followed
• what the company learned

Customer records should contain commercial context, not merely contact details and contract values.

Technical documentation should capture design rationale, not only architecture.

Postmortems should record why a failure occurred and how future operators should recognize the early signals.

Critical functions should have genuine redundancy.

Successors should practice making decisions before the original knowledge holder leaves.

This work may improve the company even if no transaction occurs.

It can shorten onboarding, improve delegation, reduce operational fragility and make management less dependent on individual memory.

In a sale process, it can also provide evidence that the buyer is acquiring an operating system rather than a collection of people held together by habit.

Some knowledge cannot, and should not, be reduced to a manual

Knowledge due diligence should not become a fantasy that every human capability can be converted into documents, embeddings and automated workflows.

Some skills remain embodied.

Some judgment develops only through repeated exposure to real situations.

Some relationships depend on personal trust that cannot be assigned through a contract.

Some organizational knowledge exists between people rather than inside any single person.

Research on knowledge transfer has found that social interaction and mentoring can materially affect whether difficult tacit knowledge is successfully transferred. The timing of transfer methods matters as well. Written documentation is only one mechanism.

There are also cultural and ethical risks.

A poorly designed knowledge-capture program can feel like surveillance.

Employees may become less candid if every conversation is treated as material for an institutional model.

Management may begin to value only knowledge that can be easily recorded, while neglecting craft, trust and practical experience.

Documentation can preserve outdated assumptions.

AI can create an illusion of understanding by producing confident summaries of a system it has never operated.

The objective should therefore not be to make the company independent of people.

Companies are not independent of people.

The objective is to understand exactly which value depends on which human capabilities, then manage that dependency consciously.

Knowledge due diligence is not a plan to replace people.

It is a plan to stop pretending that their knowledge is already owned by the company.

The next standard diligence workstream

Acquirers have become progressively better at inspecting financial statements, contracts, markets, software and cybersecurity.

The next step is to inspect the company’s ability to know.

Before paying for future cash flows, the buyer should ask:

• What does this organization need to know tomorrow in order to produce those cash flows?
• Where does that knowledge live today?
• How many people can apply it?
• What evidence demonstrates that they can?
• How long would it take to transfer?
• What happens if the current holder leaves?
• Which part can be documented?
• Which part requires observation, practice or continued relationships?
• What will the buyer accidentally destroy by standardizing the company too quickly?

These questions should not wait until the integration team begins work after closing.

They affect whether the acquisition thesis is valid in the first place.

Every transaction buys a legal company and a cognitive company.

The legal company is transferred through the purchase agreement.

The cognitive company is transferred through people, evidence, practice, relationships and time.

A serious buyer must diligence both.

The most dangerous asset in an acquisition is not the one missing from the data room.

It is the one everyone assumes is already inside it.

Sources and further reading

• Ikujiro Nonaka, “A Dynamic Theory of Organizational Knowledge Creation”, Organization Science, 1994.
• Linda Argote and Paul Ingram, “Knowledge Transfer: A Basis for Competitive Advantage in Firms”, Organizational Behavior and Human Decision Processes, 2000.
• Walter Swap, Dorothy Leonard, Mimi Shields and Lisa Abrams, “Using Mentoring and Storytelling to Transfer Knowledge in the Workplace”, Journal of Management Information Systems, 2001.
• Gabriel Szulanski, Dimo Ringov and Robert J. Jensen, “Overcoming Stickiness: How the Timing of Knowledge Transfer Methods Affects Transfer Difficulty”, Organization Science, 2016.
• Gianlucca Zuin, Saulo Mastelini, Túlio Loures and Adriano Veloso, “Leveraging Large Language Models for Tacit Knowledge Discovery in Organizational Contexts”, preprint, 2025.
• Mina Cho, Russell Funk, Alok Gupta and Mochen Yang, “Large Language Models Explain Experts Better Than Experts Themselves”, preprint, 2026.
• IFRS Foundation, IFRS 3: Business Combinations and IAS 38: Intangible Assets.

This essay is for informational purposes and does not constitute legal, accounting or investment advice.