The behaviour most consistently observed in a portfolio review meeting is that the addition of a new product is debated at length while the removal of an existing one never reaches the agenda. The proposal for a new product arrives prepared: addressable market, target margin, development calendar, and first-year unit forecast assembled into a single deck. The fate of a product that has declined in unit terms for three consecutive years, absorbed a rising service burden, and accumulated an ageing spare-parts position is, by contrast, rarely converted into an agenda item at all — not because anyone has decided to keep it, but because no trigger has been defined that would place it in front of the group. When a participant does raise its status, the answer that follows is typically relational rather than quantitative: a particular customer still orders it, a particular dealer insists on it, volume continues to arrive. The product thereby ceases to be a subject of decision and becomes a habit.
A second observation is that the same pattern operates in reverse at the product's birth. A new product is treated as having crossed a threshold the moment it moves from pilot into serial production, yet in most companies that crossing is announced not by a formal approval but by the arrival of the first large order — which is to say the stage change is declared by the customer rather than by the company. The consequence is that the product never formally enters the cost discipline appropriate to maturity: second-source qualification, revision control, service documentation, warranty provisioning. Commercially the product has matured; managerially it remains under a prototype regime, and the gap between those two states tends to surface years later as a field-failure rate or as revision chaos on the shop floor.
The mechanism underlying this behaviour is the asymmetric distribution of decision costs. Proposing a new product produces a return that is visible, dated, and attributable to a named individual; retiring a product produces a return that is diffuse, delayed, and largely anonymous, while its cost is immediate and personally assigned — a customer will be disappointed, a dealer will object, an inventory position will be written down. Under these conditions the rational manager reduces personal exposure by deferring the discontinuation decision, and the difficulty lies not in the deferral itself but in the fact that the deferral is never recorded as a decision. An unrecorded deferral does not return to the agenda six months later, because nothing exists to carry it there. The unmanaged product lifecycle is therefore not an oversight but the predictable output of the incentive structure in place.
The second layer of the mechanism is the qualitative treatment of stage definitions. Introduction, growth, maturity, and decline are conceptually understood in almost every company; what is generally absent is the numeric threshold that determines which stage a given product currently occupies — how many consecutive quarters of contraction, which gross margin level, which density of service calls, which inventory turn. Where the threshold is unwritten, stage assignment becomes a matter of interpretation, and interpretation defaults to the most senior institutional memory in the room, which in owner-managed companies is typically the founder. In the short run this arrangement is functional; the founder genuinely knows the portfolio and the resulting calls are frequently accurate. The difficulty is not accuracy but the fact that accuracy, held in a single memory, cannot be reproduced by the organisation.
On the balance sheet this configuration registers not in the product line but in the age distribution of inventory and in the provision for slow-moving stock. A product whose retirement has never been decided continues, for as long as it remains in the catalogue, to generate spare-parts obligations, minimum-order-quantity commitments, and shelf allocation; each of these is individually modest, yet across a portfolio they extend the working capital cycle without ever announcing themselves. Where inventory turns fall below the sector band during a review, the cause is almost never a single poor product but a tail that was never closed. That same tail reappears elsewhere as changeover losses in production, as part-number proliferation in procurement, and as price-list complexity on the commercial side.
The second cost appears on the measurement dimension and attaches directly to the credibility of the revenue forecast. The review table typically requests not consolidated turnover but unit volumes, average selling prices, and gross margin series at product or product-family level, usually across three years. Where the company cannot produce that series — commonly because product codes have been reassigned repeatedly in the ERP, because variants have been split across separate items, or because margin is calculated only at the consolidated level — the source of historical growth remains unverifiable. Growth that cannot be traced to a product is not treated as repeatable in the model, and the forecast is pulled toward the most conservative scenario available. The valuation effect arises here not as a penalty but as the mechanical consequence of absent evidence.
The third cost consolidates the ownership and continuity dimensions. Where no accountable owner is defined for the lifecycle, new-product decisions are driven by commercial pressure and discontinuation decisions by whatever constraint is binding in production or procurement that quarter; the portfolio is shaped, in other words, not by portfolio strategy but by the loudest constraint of the period. Review teams rarely ask about this directly. They ask instead which products have left the catalogue in the past two years and where those decisions were recorded. In a company holding no discontinuation record, the continuity dimension cannot be verified, and that gap tends to migrate into the closing structure as an earn-out indexed to product revenue or as an expanded set of representations and warranties.
The structural intervention is built through decision architecture rather than individual awareness, and it separates into four components. The first is the numeric definition of stage-transition thresholds: the criterion declaring that a product has moved from growth to maturity, or from maturity to decline, is written in terms that leave no room for negotiation, and any product reaching the threshold enters the agenda automatically. The second is a mandatory three-option decision set for every product that enters the agenda — renewal investment, price and channel repositioning, or a discontinuation calendar — with deferral admissible only where a rationale and a review date are recorded alongside it. The third is that the decision is logged at the moment of proposal rather than the moment of approval; a record captured at approval shows neither who argued what nor on which data, and leaves nothing to be learned a year later. The fourth is that a discontinuation decision attaches to a closure list rather than a date: last order date, spare-parts commitment horizon, service support window, inventory run-down plan, and customer migration path.
BEIREK generally constructs this intervention across two layers. In the first, the existing portfolio is decomposed into a unit-price-margin series at product level, and each product is assigned to a stage against the defined thresholds; the output of that exercise is not a report but a table the company can regenerate periodically from its own systems, since a one-off decomposition carries no evidentiary value at the review table. In the second layer, the cadence through which that table converts into decisions is put into operation: a portfolio review session on a fixed calendar, defined participant roles — product owner, production or operations, procurement, finance — and a decision log closed at the end of each session against the three-option set.
The most tangible output of that cadence in its early months is usually not a new product decision but the closing of a tail that no one has owned for years; the catalogue narrows, part-number diversity falls, and the age distribution of inventory improves. The second output, equally consequential, sits on the documentation dimension. A decision log operated for six months accomplishes in a review process what a policy document cannot, because it demonstrates that the practice actually runs. A procedure the company has written for itself is a statement of intent; a series in which decisions are recorded together with date, rationale, and outcome is verifiable evidence of behaviour, and review teams consistently prefer the second to the first.
The continuity dimension is tested in the first session held without the founder in the room. Where thresholds are written, the data set is generated from the system, and the decision set is mandatory, that session closes without recourse to founder judgment; the quality of the resulting decision will probably be lower on the first attempt, but the mechanism has operated and can be repeated. This is precisely what the investor is looking for — not the present accuracy of the portfolio, but a demonstration that the accuracy can be reproduced by the company. Where that demonstration is made, the negotiating force behind contingent consideration structures tied to product revenue weakens appreciably, since such structures are, in substance, the pricing of repeatability that could not be proven.
The measure of portfolio management is not how many products sit in the catalogue but how many were closed, with a stated rationale, over the past two years. Where that number is zero, the company is not managing its portfolio but carrying it — and every product carried in this manner waits as a file that someone else will eventually open.