Business Model Assessment

Why this node matters

Most Mittelstand portcos treat the business model as fixed. A strategic question, decided at founding, too slow to move within a hold period. PE operating practice has historically reinforced this, leaving the higher-impact design theme unutilised.

The unexamined assumption underneath: that the architecture set at founding is the architecture that maximises value at exit. The assumption is rarely tested. The real differentiator is the gap between the model the portco runs and the model its customer capital could be monetised through. That gap is what this node has to find and close.

Architecture, not effort, sets the ceiling. PE operating practice treats customer capital as a customer success problem. But customer success operates the asset; the business model creates it. Reinartz and Kumar (2002), studying 16,000 customers, found loyalty and profitability correlate below 0.5. Effort against the relationship does not produce financial value unless the architecture monetises it. McKinsey (2025) found best-in-class NRR practices in B2B drive 16 percentage points of uplift, enabled by architecture rather than effort. The lever is structural, not operational.

The model is a measurable variable, not a fixed constant. Sohl, Vroom, and Fitza (2020), analysing 917 European retailers over twelve years, found business model explains 5.1 percent of variance in ROA and 7.9 percent of variance in market share: a magnitude comparable to industry effects. Zott and Amit (2007), studying 190 firms, showed novelty-centred design predicts performance independent of strategy or industry. Most Mittelstand portcos run efficiency-centred models because they were built thirty to fifty years ago, not because the design was tested and chosen. A variable that explains performance at the magnitude of industry itself is being treated as if it were not a variable at all.

Partial redesign is achievable within a single hold. The objection to revisiting the model is that it is too slow. The objection does not survive contact with the operational evidence. Partial redesign is achievable in 12 to 18 months: adding a recurring-revenue layer, restructuring payment terms, outcome-based contracts for the top decile, sale-to-lease for the largest SKUs. Each produces measurable EBITDA and customer capital effects within a single hold. The architecture does not have to be rebuilt to move; it has to be moved at the decile and layer where the customer capital already sits.

Legibility of customer capital is what the multiple pays for. Aventis Advisors (2025), analysing 1,325 software transactions, found recurring-revenue models command 2 to 5x the EBITDA multiples of transactional models for otherwise identical performance. The premium is not for the revenue; it is for the legibility of customer capital to an acquirer. A transactional model holds the same customers but cannot show them on the balance sheet of the deal. The business model is the instrument that converts a relationship an acquirer cannot price into an asset an acquirer will pay a multiple for.

When a portco cannot answer how much of its customer capital is structural versus operational (held by the architecture versus held by the effort of its people) then it is managing the revenue it built rather than the revenue its model could capture, and the retention and expansion lines of the EBITDA bridge are resting on an architecture that has never been examined.

The thesis: a business model that has not been revisited since founding is not a strategic choice. It is an inherited constraint.

Revenue, cost, and margin drivers

• Revenue model choice sets the multiple. Median software multiples sit at 3.0x EV/Revenue and 15.2x EV/EBITDA, against 1.4x and 11.0x for hardware and 1.3x and 10.2x for IT services (Aventis Advisors, 2025). The gap is driven by recurring revenue and gross margin structure, not by industry.
• Scope of value capture is a model decision, not a sales one. A portco selling industrial equipment may have a customer paying for equipment, installation, training, maintenance, consumables, and certification (Christensen, Hall, Dillon, and Duncan, 2016). A transaction-revenue model captures one of the six. A recurring service-and-platform model captures five.
• Pricing logic is constrained by architecture. Business model architecture determines what pricing logic is available (Nagle and Müller, 2018). Cost-plus is the default in product-led models. Value-based pricing requires subscription, outcome-based, or tiered architecture. The 8 to 11 percent EBITDA effect from each 1 percent of price realisation (Marn and Rosiello, 1992) is structurally capped by the model.
• Working capital is a model property. Subscription models run on negative working capital: customers pay upfront, suppliers pay later. PE buyout multiples cluster around 11x EBITDA for technology with strong working-capital dynamics versus 6 to 8x for cyclical industrials (McKinsey, 2024).
• Transaction-cost burden is releasable cost. Mittelstand portcos with complex channel structures, customised offerings, and per-customer pricing often have transaction costs running 15 to 25 percent of revenue (Williamson, 1985; Zott and Amit, 2007). Redesign that simplifies these flows releases material cost.
• Marginal cost of revenue sets the margin ceiling. Subscription architectures have near-zero marginal cost per additional unit; transactional architectures do not. This, not operational discipline, sets the ceiling.
• Retention economics only compound under the right architecture. Reichheld and Sasser (1990) established that a 5 percent retention improvement drives a 25 to 95 percent profit increase, structurally enabled by recurring-revenue mechanics. McKinsey (2025) extends this for B2B SaaS: 16 percentage points of NRR uplift compounds into 5 to 10 EBITDA multiples of valuation difference.
• Customer concentration is a multiple-compression factor. Diversified customer bases produce structurally higher and more durable margins than concentrated ones; concentration compresses the multiple at exit (First Page Sage, 2025).
• Architecture coherence outpredicts any single lever. Sohl and Vroom (2014) found business model coherence predicted long-run margin more strongly than any single design element. Internally consistent design choices, not the strongest individual lever, predict the margin ceiling.