In the organisation session of an investment review, when the org chart uploaded to the data room is placed beside the notes of three separate management interviews conducted within the same week, a recurring angular difference becomes visible: two functions shown as distinct on the chart converge, in daily practice, on a single desk; a decision that appears on the chart to belong to a named unit does not in practice advance without the founder's approval; and an entire body of work absent from the chart altogether — closing out a customer complaint, absorbing a supplier delay, approving a pricing exception, which is to say precisely the work that determines margin — cannot be attributed with any clarity to a defined position by anyone in the room. This divergence rarely originates in bad faith or in neglect. It originates in a growth rate that has outpaced the rate at which the description of the work is refreshed.
In most companies job descriptions are written twice, at two different moments and for two different purposes, and this duality is the source of everything that follows. The first drafting occurs at the point of hiring, its purpose being to communicate to a candidate what is expected of the role; the resulting text sits close to advertisement language, arrives as a list of verbs, and rests on the market's generic characterisation of the position rather than on the company's actual decision flow. The second drafting occurs because of an audit, a certification or a customer requirement, its purpose being to satisfy an external party's checklist; that text looks more formal but is bound even more loosely to practice, having been produced retrospectively from an assumed structure rather than from observation of what the incumbents actually do. Both share the same omission: neither states the measure against which the output of the role is assessed, nor the person from whom account is taken when that output fails.
The underlying mechanism arises from an asymmetry between how organisations describe work and how they distribute it. Description is slow and costly, requiring drafting, alignment, approval and periodic revision. Distribution, by contrast, is instantaneous and almost always verbal, governed at the moment of allocation by who is available, who is faster and who is least likely to object. In the short run this preference is entirely rational, since getting the work done is more urgent than describing it; the difficulty appears when the condition changes — when the team grows, when a second location opens, when the founder's available hours thin out — and the distribution reflex remains constant while description never catches up. The organisation then begins operating within a widening interval between its written structure and its functioning one, an interval that becomes invisible from the inside precisely because everyone working there already knows who does what.
That form of knowing is a product of personal rather than institutional memory, and its non-transferability is what makes it problematic under review. A buyer, an investment committee or a credit committee can already observe that the company works today; that is not what is being sought. What is sought is a demonstration that the same work can be performed to the same standard not by the current roster but by tomorrow's. At this point the job description ceases to be a human resources artefact and becomes evidence about the source of performance: where the scope, the measure and the owner of a task are written down, and where the correspondence between that written form and actual practice can be shown through records, the source of performance is process; where it cannot be shown, the source of performance is a person, and a person does not transfer with the shares.
The institutional cost accumulates first through the ownership gap. The layer most frequently omitted from job descriptions is not what the work is but where the boundary of decision authority over that work ends — up to what amount a discount may be approved, which delay may be communicated to a customer unilaterally, which supplier substitution requires a second signature. Absent written boundaries, the employee selects the safe course and escalates, and every escalated decision occupies a place on the founder's agenda; as that agenda fills, decision latency lengthens. The resulting picture is read in diligence through a small set of indicators — proposal turnaround, order approval cycle, complaint closure time — and the fact that all of them converge on the same individual is recorded as the quantitative evidence of key-person dependency.
The second cost channel sits on the measurement side and erodes predictability directly. Where a description remains a list of verbs, the position generates no data indicating whether it is performing well or poorly; appraisal rests on impressions, compensation is settled by negotiation, and when someone departs there is no basis for estimating how long a replacement will take to reach equivalent output. For an investor the consequence is that the growth assumption in the business plan cannot be tested on the human resource side: how many people a doubling of revenue requires, how long those people take to become productive, and at what point the existing team reaches its capacity ceiling are questions that remain unanswerable. An organisation that cannot be measured is a cost structure that cannot be modelled, and a cost structure that cannot be modelled is priced, invariably, from the conservative end.
The third channel surfaces in transaction structure and typically appears in terms before it appears in price. Where the descriptions are found not to match actual practice, the buyer's standard reflex is not to cut the headline number but to spread the risk across time: retention covenants requiring key personnel to remain for a defined period, a post-closing service obligation on the founder, earn-out tranches conditioned on continuity of performance, clauses converting completion of organisational documentation into a condition precedent, and definitions treating the departure of named individuals as a trigger event. Each of these represents, for the seller, consideration that is both deferred and contingent; even where the nominal valuation appears preserved, the difference between the present value of that consideration and its risk-adjusted equivalent frequently exceeds what a straightforward discount would have cost.
The first component of a structural intervention is to record how the organisation actually works before writing any description of it. A workable job description is derived not from what the position ought to do but from the decisions the incumbent has in fact taken over the preceding quarter, the approvals sought, and the tasks handed on to others; the starting point is therefore an inventory of live decisions rather than a blank template. The second component is closing every description with four fixed fields: the scope of the work, the monetary and operational limit of decision authority, two or three indicators against which output is measured, and the deputising line identifying who steps in when the incumbent is unavailable. The third component is a revision rhythm, with descriptions updated not through an annual ceremony but against events that materially alter the organisation — a new customer segment, a new site, headcount crossing a defined threshold, or a key position changing hands.
BEIREK conducts this work as an exercise in decision architecture rather than in human resources documentation. Our practice begins with a decision inventory: the recurring decisions bearing on cash, on schedule or on customer relationships are listed, and for each the person who in fact proposes, the person who approves and the person held accountable for the outcome are marked separately. The rows in which those three roles collapse into one individual, or in which none of them can be located, produce the real map of the definitional gap. Descriptions for each position are then written backwards from that inventory, with authority limits committed to writing, so that the description leaves the language of a job advertisement and becomes an instrument of delegated authority.
The second stage establishes the evidence chain that prevents the description from remaining a paper exercise. Each description is bound to a regular output the position produces — a report, an approval record, a review minute — and the existence of that output is verified within a monthly management rhythm; what enters the data room at diligence is therefore not the text of the description but the series of records demonstrating that it has operated for twelve months. Within the same rhythm the deputising line is tested, planned absences of key positions being used to observe whether decisions genuinely progress under the named alternate; where they do not, the gap lies in authority rather than in description, and is corrected there. Once both layers are in place, the continuity question ceases to be an assertion and becomes a verifiable record.
What distinguishes job descriptions from the other headings under organisation and management structure is that remediation requires time rather than capital, and once the diligence calendar has begun the remaining time is generally insufficient. Where a full set of descriptions is prepared after a transaction has been announced, every document carries the same approval date, and that simultaneity tells an experienced reviewer that the descriptions were generated by the transaction rather than by the operation. The same set, presented with eighteen months of successive versions, the change notes explaining each revision, and the regular output records attached to the roles, proves something different altogether: that the company possesses the capacity to observe and describe its own functioning.
Ultimately, what an investor looks for in job descriptions is not who does what; that much is learned in a handful of interviews. What is sought is whether the company knows this on its own account, whether it can demonstrate that it knows, and whether the knowledge resides in the institution or only in the memory of a few individuals. The distinction that determines a company's valuation lies less often in the magnitude of performance than in the demonstrability of its repeatability once the individuals change; and job descriptions remain the least expensive, and most frequently deferred, instrument of that demonstration.