In an investment committee session, the growth narrative is typically constructed across three curves: monthly new customer count, aggregate transaction volume, and the compound growth rate of gross revenue. To the extent that discussion concentrates on the slope of those curves, another magnitude tends to pass through the same deck on a single slide and as a single average figure — what one customer actually costs the company. That figure is almost always bounded by direct cost, while acquisition spend, payment processing fees, the burden of returns and rework, site visits, call centre contacts, and the labour equivalent of support tickets either sit aggregated under operating expenses or have never been separated at all. The arithmetic of the presentation remains internally consistent in this state; what is missing is not consistency but the unit itself.
A second and less frequently noticed observation is that the person who actually knows this figure inside the company is generally not the person presenting it. True cost per order is known to the warehouse or operations manager; true service load per contract is known to whoever schedules the field team; and that knowledge lives not in the financial reporting system but in an operational tracking sheet, usually broken down along lines that do not map onto the chart of accounts. Where the two systems remain unconnected, the organisation possesses information about unit cost without possessing an institutional record of it — and what sits on the table at the moment of decision is the record, not the information. This disconnection does not, on its own, constitute a management weakness, given that the two systems were built for different purposes; what determines the outcome is that responsibility for the bridge between them was never assigned.
The name for this pattern is unit-economics failure — revenue per customer, transaction, or contract falling short of the fully attributed cost of that unit — and its mechanics operate on two levels. The first is definitional: what counts as the unit looks like a measurement preference but functions as a decision. Within the same company, defining the unit as the customer can produce a positive picture while defining it as the order produces a negative one, since the first definition admits the repeat purchases a customer will make over their lifetime and the second does not. The second level is allocative: whether a given cost is variable or fixed depends on the volume range under examination, and where warehouse rent sits beside courier fees in the same table, unit margin cannot be computed until the question of which item scales with volume has been answered.
This tendency warrants reading not as an error but as a rational choice under specific conditions. An early-stage company losing money on the first transaction may be purchasing the learning curve, building density economics, or establishing a habit; and where retention is sufficiently high, the negative margin on that first transaction becomes the entry ticket to a cohort that turns positive on the third or fourth. Likewise, in a fixed-cost-heavy structure, declining unit cost as capacity utilisation rises is an expected outcome, and selling below cost while waiting for that decline is a defensible strategy. For the choice to retain its character as an investment, one condition applies: the expiry date and the success criterion must be written down at the outset.
The moment the choice converts into a cost is the moment those two elements remain undefined. Where no expiry date is written for the subsidy, the promotional price becomes the reference price over time; the customer builds their own cost structure around it, sales incentive compensation is calculated against it, and raising the price ceases to be a commercial decision and becomes a negotiation that puts part of the existing revenue base at risk. Where no success criterion is defined, scale itself begins to be read as evidence of improvement — yet scale reduces unit cost only where the dominant cost item is fixed in nature. Where the dominant item is a variable cost priced by a third party — delivery fees, payment processing, cloud consumption, field service — volume growth does not compress unit cost; it simply enlarges the multiplier on the same gap.
The balance-sheet expression of this structure typically appears not in the profit and loss statement but in the working capital cycle. In a company growing on a negative unit margin, revenue growth triggers cash outflow ahead of cash inflow; where suppliers are paid in advance and customers settle on terms, the growth rate and the burn rate move in the same direction, and the company finds itself financing its own success from outside. Funding need at that point is no longer growth capital but gap-closing capital; to the extent that both are narrated in the same deck under the same heading, the picture seen by the capital provider and the picture lived inside the company diverge. On the credit side, that divergence shows up in working capital covenant headings and in the conditions attached to drawdown.
The valuation consequence is sharper and generally arrives in a single move: a company discussed on a revenue multiple becomes, the moment unit margin is interrogated, a company discussed on a contribution-margin multiple. The difference between those two bases is, as a matter of order of magnitude, more determinative than any negotiation over the multiple itself, since once part of the revenue base is accepted as permanently margin-free, the debate migrates from what multiple applies to which revenue counts. What the counterparty typically requests at that stage is not an opinion but a table: contribution margin net of acquisition cost, tracked over time on a monthly cohort basis.
At the diligence table, the primary difficulty concerns the existence of that table before its contents. Where cohort-level contribution margin has not been recorded contemporaneously, any version reconstructed within the closing timetable is necessarily assumption-laden — and an assumption-laden table registers with the counterparty as a risk item rather than as evidence. The pricing mechanisms for that risk item are well established: a direct discount, an earn-out conditioned on verification of cohort behaviour, or an elevated escrow ratio accompanying an expanded set of representations and warranties. What the three share is that the cost arises not from the company's actual performance but from the gap in that performance's demonstrability.
The mechanism that neutralises this tendency is institutional architecture rather than individual attention, and it separates into four components. The first is definition governance: what constitutes the unit, which costs are attributed to it, and who holds authority to alter that definition are fixed in a single document, with any change processed as a recorded decision rather than a reporting detail. The second is full cost attribution: acquisition spend, payment infrastructure, returns and rework, support contacts, and field intervention enter the unit table instead of remaining pooled under operating expenses. The third is measurement rhythm: cohort contribution margin is reported on a calendar independent of sales volume and in a consistent format. The fourth is subsidy discipline: every segment operated at a negative margin carries a termination date, a return criterion, and a named owner.
BEIREK operates this structure as an extension of the decision architecture it builds for capital-intensive, financed projects. In the engagements we run, the unit definition is fixed once according to the economic logic of the project or business line — per installed megawatt, per contract, per active subscriber-month — and carried as a single reference across the financial model, operational reporting, and lender reporting; where three surfaces hold three different unit definitions, the argument over which set of numbers is correct predictably slows the pace of decision. We construct the cost attribution matrix jointly with the operations team, documenting which item behaves as variable within which volume range rather than leaving that judgment to be improvised under diligence pressure.
The second line of intervention concerns rhythm. For every segment operated at a negative margin, a termination date, a return criterion, and an owner are entered into the record; that entry opens at the moment of proposal rather than at the moment of approval, because the rationale for a subsidy decision can be written most cleanly when it is taken, not after the outcome has become visible. Fixed-interval review sessions address only two questions: whether the criterion has been met, and if not, whether the period is being extended or closed. Once the record of those two questions accumulates, the company enters pre-closing diligence with a chain of decisions documented in their own time rather than with a table reconstructed after the fact, and the uncertainty the counterparty prices narrows in proportion to the length of that chain.
Unit economics is the only measure that captures not how quickly a company is growing but what growth is doing to the company. A negative reading is not in itself the problem; the problem is the absence of institutional knowledge as to whether that negativity is a choice or a consequence. The distinction between the two is settled not by the persuasive force of the presentation but by the record kept on the day the decision was made.