In a compensation review meeting, when someone asks why two employees carrying the same title are paid differently, the answer rarely turns on performance, tenure, or the scope of responsibility; it turns instead on the dates the two joined and on the negotiating conditions that prevailed on each of those dates. The first was hired during a period of constrained cash flow, anchored to a low base; the second arrived under the pressure of a delivery deadline, in circumstances that required the offer to close within a week. Three years later the gap has become impossible to explain on institutional grounds and equally impossible to close cheaply, since aligning upward raises total personnel cost permanently and in a single step. The gap is therefore preserved, new hires are layered on top of it, and the company arrives — through a sequence of individually defensible decisions — at a pay structure it can no longer explain to itself. The distinguishing feature of this accumulation is that no single decision in it was wrong; each was defensible at the moment it was taken, and none was connected to the next.
At the diligence table, the first place this configuration becomes visible is not the salary levels themselves. When the investor side asks for compensation bands, what typically arrives is a payroll extract listing title, gross salary, and hire date — a document that answers a different question from the one that was asked, because the request was never for the list but for the rule that produced it. This is precisely the distinction the existence dimension tests: a payroll extract shows what compensation is, whereas a pay architecture shows what the next compensation decision will be. For the party conducting the review, the distance between those two documents is the same distance that separates historical performance from repeatable capacity, and valuation is determined almost entirely by the second.
The mechanism beneath this configuration is not a management failing but a shortcut calibrated to a particular scale. In a forty-person organization, a compensation decision is made by a founder or general manager who knows the individual personally, who treats the candidate's current salary as an anchor, and who reads the counterparty's resistance intuitively; the method is fast, it avoids the fixed cost of building a system, and at small scale it converges reasonably well on the right answer. The problem lies not in the shortcut but in its persistence after the underlying conditions have changed: once headcount reaches two hundred, hiring runs at several offers a month, and three separate people are making offers, the same intuitive method stops producing consistent output, because intuition is person-specific and cannot be delegated. From that point forward, the company is no longer managing compensation decisions; it is managing the inconsistencies among them.
On the variable pay side, the same mechanism produces a sharper result. Bonuses in most companies are determined at year end, somewhere between realized profit and management discretion; a formula exists but is unwritten, or is written but has not been applied for two cycles, or has been applied while the calculation trail resides in one person's spreadsheet. Running alongside this, verbal commitments accumulate — equity, profit participation, exit bonuses extended to retain a critical employee — none of which ever reached contractual form, which means none appears in the accounts while all of them sit intact in the employee's memory and expectation. What the documentation dimension tests is the asymmetry here: an unverifiable commitment cannot simply be disregarded by an acquirer, because the obligation side of it is real; it is merely unquantified.
The channels through which this gap reaches the balance sheet and the transaction structure are reasonably well defined. The first is earnings normalization: founder compensation running materially below or above market, family members carried on payroll, and irregularly paid bonuses are all restated as adjustment items in the buyer's model, and because that restatement changes the earnings base before the multiple is applied, its effect passes into price on a leveraged basis. The second is that the dispersion within the pay structure generates a cost forecast of its own: where a single role carries a wide salary range, post-acquisition alignment is modeled by the buyer not as a one-time expense but as a permanent upward shift in the personnel cost line. The third is the migration of unwritten commitments into representations and warranties, pre-closing conditions, and the escrow percentage; an undocumented equity promise tends not to reduce the headline price so much as to withhold a portion of it for a defined period after closing.
The measurement dimension is where most companies are weakest, because compensation is tracked as an expense line rather than managed as a discipline. Total personnel cost as a share of revenue is monitored, while the indicators that would actually inform decisions — where each role sits relative to band midpoint, the acceptance rate on offers extended, the relationship between salary increases and voluntary turnover, unit cost by role — are not produced on any regular cycle. The consequence is that salary increases function less as a performance mechanism than as a response to a resignation signal, and the observable evidence of that reactivity is that increases cluster not around the review calendar but around the resignation letters of critical staff. Without measurement, management learns whether the compensation policy works only at the moment it stops working, which is to say after the loss has already occurred.
Under the ownership dimension, the operative question is not who makes the compensation decision but who is able to refuse one. In many mid-sized companies the human resources function administers the compensation process without holding decision authority; when an out-of-band offer arises, approval escalates to a single individual, typically the founder, and the exception that individual grants becomes the reference point for the next negotiation. Where authority is not distributed, policy erodes at the speed of its exceptions. The continuity dimension carries the picture one step further and asks whether, if the founder remained outside the decision process for six months, compensation offers would still be produced consistently or the process would simply stall. For an investor, the answer to that question is considerably more binding than the salary levels themselves, since the transferability of the compensation decision is among the most concrete places founder dependency can be measured.
The mechanism that neutralizes this tendency is institutional architecture rather than individual discipline, and it decomposes into four separable components. The first is role architecture: levels defined not by title but by scope of responsibility, decision authority, and required depth of expertise, with a salary range attached to each level. The second is the decision record, and its timing is the critical variable; the record is kept at the moment the offer is constructed rather than at the moment it is approved, because a record created at approval documents the outcome and not the reasoning. The third is the authority matrix, specifying in advance which level closes in-band offers, what threshold triggers which approval for exceptions, and how frequently granted exceptions are revisited. The fourth is the measurement set: position relative to band midpoint, dispersion width by role, offer acceptance rate, and the calendar clustering of increases, produced with the same regularity as operating indicators.
The intervention required on the variable pay side is not a more elaborate formula but a visible calculation trail. A written bonus plan is not sufficient on its own; what must be traceable is which data feed the plan, in which system that data resides, who performs the calculation, and why amounts actually paid over the last three cycles diverged from the amounts the plan would have produced. Where that traceability exists, variable pay ceases to be an uncertainty item in diligence and becomes evidence of the relationship the company has established between performance and payment; where it does not, the same line is typically modeled by the buyer under its most conservative scenario.
BEIREK's work in this area neither begins nor ends with the delivery of a compensation policy document. In practice the existing salary distribution is first mapped against role levels, the differences that have formed within a single level are separated according to whether they originate in scope or in hire date, and for the portion that genuinely requires correction a convergence plan spread across successive review cycles is constructed rather than a one-time alignment; the effect on personnel cost thereby becomes something that can be modeled in advance. The offer construction flow is then rebuilt: band definitions, a reasoning record captured at the moment the offer is formed, a defined threshold and approval level for out-of-band exceptions, and a rhythm under which those exceptions are reviewed periodically.
The second workstream concerns demonstrating that the structure operates independently of the founder, since what persuades in a diligence process is not the existence of a policy but the evidence that the policy produced the same outcome while the founder was outside the room. This requires that compensation decisions have been taken within the defined authority matrix over a meaningful period, that the record of those decisions remains accessible, and that the measurement set has been produced across at least two review cycles. On the verbal commitments side, the work consists of identifying existing expectations, converting into contractual form those that can be converted, and placing on record that those which cannot are excluded from scope; this remains among the small number of preparatory measures that measurably reduce pressure on escrow and warranty coverage during closing negotiations.
Compensation is the line item that reveals not how much a company pays its people but the rule by which it makes its own decisions, which is why the weight it carries at the diligence table far exceeds its share of the expense base. A buyer can model high salary levels and reflect them in price; what a buyer cannot model is a company that does not itself know why its salary levels are where they are — and whatever cannot be modeled is priced, predictably, through discount or through withholding.