In a company with two or three founders, there is a moment that repeats itself several times a month and is almost never recorded: a decision arrives that does not clearly belong to anyone, and it is resolved in a corridor, a message thread, or a five-minute exchange before another meeting. A pricing exception for a strategic customer, a hiring decision one level below the leadership team, a supplier payment term that deviates from the standard, a scope change on a project already in delivery. The decision gets made, usually well, and the company moves on. What does not happen is any record of who held the authority, what alternatives were weighed, or what would happen if that same question arrived while one of the founders was unavailable for two weeks.
Observed from the outside, the striking feature of this pattern is not disorder but its opposite. Founders who have worked together for years develop a settlement so efficient that it needs no articulation: each knows which questions the other will take, which ones require a joint decision, and which ones are simply not worth raising. The arrangement is fast, low-friction, and generally correct. It is also entirely undocumented, and that combination — high functional quality paired with zero external verifiability — is precisely what makes it difficult during an investment review.
The mechanism sustaining this is not negligence. Formalizing a working arrangement carries an immediate and visible cost — the time to draft it, the discomfort of naming boundaries between people who trust each other, the risk of appearing to distrust a partner by asking for written authority limits — while its benefit is deferred, probabilistic, and invisible until the moment it is needed. Under those conditions, deferring formalization is a rational allocation of scarce attention. The difficulty is that the calculation does not update when the company's conditions change: the arrangement designed for two founders and eleven employees continues unmodified at seventy employees, three business lines, and a leadership layer that now has to guess which founder to approach.
A second mechanism reinforces the first. Because the founders resolve ambiguity in real time, the organization never experiences the cost of the ambiguity — the founders absorb it. Every unclaimed decision is quietly caught, and the catching is invisible in every management report the company produces. The absorbed cost only becomes measurable when the absorbers are removed, which in a transaction context is exactly the scenario the investor is modeling.
This is why the review does not ask whether the founders divide responsibility effectively. It asks six narrower questions, and each addresses a different failure mode. Whether the division exists as a defined structure rather than a claim made in a management presentation. Whether it is documented in current, approved, retrievable form — a board-approved authority matrix, signed job descriptions, a delegation-of-authority schedule with monetary thresholds. Whether daily practice actually matches the document, or whether the document was produced for the data room three weeks before the process opened. Whether anything about it is measured, meaning whether decision cycle times, escalation frequency, or approval throughput are visible anywhere. Whether each domain has a named owner with defined decision rights and an accountability path. And whether the whole structure continues to function when a founder is absent for a quarter.
The last two dimensions carry disproportionate weight, because they are where founder dependency becomes legible. An investor reading an organizational chart is looking for the zones that belong to no one — not the areas of overlap, which are visible and usually harmless, but the unclaimed spaces between founders where decisions wait for an informal resolution that only the founders can supply. Those zones are where post-closing delay concentrates, and where an acquirer's integration plan tends to fail on schedule rather than on substance.
The cost surfaces on specific commercial surfaces rather than in general concern. When authority boundaries cannot be evidenced, the reserved matters schedule expands, because the investor compensates for undefined internal authority by pulling decisions upward into the shareholders' agreement — which slows the company precisely in the areas where founder speed had been an asset. Warranty coverage widens to include representations about management authority and the absence of undisclosed commitments, since an undocumented delegation structure makes it harder to confirm that no one bound the company outside their remit. Escrow ratios and survival periods tend to move in the same direction for the same reason.
The most consequential effect is on transaction structure itself. Where the review concludes that performance is attributable to a division of labor that exists only between two individuals, consideration shifts from upfront cash toward deferred and conditional components, with earn-out milestones tied to periods during which both founders remain in place. This is not a punitive structure; it is a rational response to an unverifiable dependency. But it transfers the timing and the risk of the founders' own institutional design failure onto the founders' proceeds, which is a price rarely anticipated at the point where formalization was postponed as unnecessary.
There is also a valuation channel that operates before any negotiation. In comparable-company reasoning, a business whose leadership structure is reproducible is treated as a platform; a business whose leadership structure is a specific pair of people is treated as a practice. The multiple applied to a platform and the multiple applied to a practice differ by an order that no amount of trading performance in a single year will close. What determines the classification is not profitability but demonstrability — whether the company can show that its results are produced by a structure rather than by two individuals who happen to work well together.
The intervention is architectural rather than personal, and it does not require the founders to change how they work. The first component is a delegation-of-authority schedule with monetary and categorical thresholds, board-approved and dated, that states which decisions each founder may take alone, which require joint agreement, and which escalate to the board. The second is a decision log maintained at the point of proposal rather than at the point of approval, which is the difference between a governance record and a formality — a log written only when decisions are approved captures outcomes, while a log written when decisions are raised captures authority, alternatives, and the reasoning that a diligence team actually needs. The third is a defined escalation path for the unclaimed zone, so that decisions belonging to no one have a stated default holder rather than an implicit one. The fourth is a review rhythm — a quarterly reconciliation between the written authority matrix and the decisions that were actually taken, which is the only mechanism that keeps the document from drifting into fiction.
In our work with founder-led companies preparing for capital raises, acquisition, or generational transfer, we begin by reconstructing the authority map from evidence rather than from interviews: the last two quarters of approvals, contract signatures, payment authorizations, and hiring decisions, mapped against who actually made each call. That reconstruction almost always produces a picture that differs from what the founders describe, and the difference itself is the finding — it identifies which domains have drifted, which have no owner, and which are held by one person without any recorded substitute. From there we install the delegation schedule, the proposal-stage decision log, and the quarterly reconciliation, and we run that rhythm long enough for it to generate its own history, because a governance document with no operating record behind it is read by a diligence team as a document produced for the diligence.
The sequencing matters more than the content. An authority matrix approved four weeks before a data room opens is worth very little; the same matrix with four quarters of decisions logged against it, including the exceptions and how they were resolved, is a different asset entirely, because it demonstrates not the intent to govern but the practice of governing. The company that begins this work when it has no transaction in view is the one that later negotiates on structure rather than on discount.
A useful way to test where a company stands is to ask what would happen to the decision queue if one founder were unreachable for a full quarter — not whether the company would survive, which it would, but which specific decisions would wait, and how long. If that list can be produced from records rather than from memory, the division of responsibility is a structure. If it can only be produced by asking the founders, it is still an arrangement between two people, and every party pricing the company will treat it accordingly.