Subscription-hardware businesses sit awkwardly between two modelling traditions. The SaaS template, refined over the past decade, treats revenue as a cohort waterfall, foregrounds annual recurring revenue and net retention, and reduces the balance sheet to a thin row of capitalised commissions and contract assets. The hardware template, inherited from manufacturing, foregrounds bill of materials, gross margin per unit, inventory turns and supplier credit. Each is internally coherent. Combined in a single business, they rarely add up.
The error is not arithmetic. It is structural. SaaS templates underweight inventory drag and supplier-terms risk because they assume a near-zero working capital cycle. Hardware templates underweight the recurring-margin uplift that the software layer contributes once a unit is installed. Reported gross margin looks unstable; cash conversion looks worse than the underlying economics warrant; and committees end up debating the wrong number.
This note argues for treating the combined revenue line as a single contribution-margin structure, supported by an explicit working-capital module that surfaces inventory, supplier terms and customer-financing dynamics as first-class assumptions.
Why the conventional templates mislead
A pure-SaaS model typically assumes that the cost of revenue is dominated by hosting and customer success, that gross margin sits between 70 and 85 per cent, and that working capital is essentially neutral once collections are stable. Deferred revenue funds part of the cash cycle. Inventory does not appear.
A pure-hardware model assumes a unit-economic build: revenue per unit, variable cost per unit, contribution per unit, with inventory days, payable days and receivable days driving the cash cycle. Recurring revenue, if it exists, is bolted on as a separate sheet.
Subscription-hardware businesses inherit the worst of both. They carry inventory (often more than a pure manufacturer, because component lead times have lengthened); they finance customer hardware acquisition (either explicitly through leasing structures or implicitly through subsidised upfront pricing); and they accrue a recurring software margin that does not show up in any conventional gross margin line.
Three failure modes follow.
- Reported gross margin oscillates with mix. A quarter heavy on new hardware shipments looks unprofitable; a quarter heavy on installed-base subscription revenue looks unrealistically strong. The committee sees volatility where there is, in fact, a stable underlying contribution structure obscured by accounting presentation.
- Working capital absorbs cash that the P&L does not explain. EBITDA grows; cash does not. The model does not surface the linkage because inventory and supplier-terms assumptions sit in a corner of the balance sheet, not in the operating logic.
- Capital structure conversations stall. Lenders ask for unit economics; the founder presents ARR. The two are talking past each other. The model does not bridge them.
The contribution-margin view
The proposal is straightforward in principle and demanding in execution. Treat each customer (or cohort, depending on the granularity the data supports) as a single revenue object generating two streams: an upfront hardware sale (or a financed equivalent) and a recurring software subscription over an estimated relationship life. Sum the cash inflows, sum the directly attributable cash outflows over the same horizon, and compute contribution at the relationship level.
The directly attributable outflows include the obvious items (bill of materials, freight, installation, payment processing, hosting attributable to the active unit, customer support) and the less obvious ones (warranty accrual, software maintenance attributable to the installed base, customer-financing cost of capital where the business carries the receivable, and any recurring component replacement cycle).
Computed this way, the gross margin trajectory is no longer a quarterly artefact. It is a function of three drivers: the hardware contribution at the point of sale, the recurring software contribution per period, and the relationship duration over which the second stream accrues. Each driver is interrogable. Committees can challenge the relationship-life assumption directly rather than arguing about whether 73 per cent gross margin is the right number.
The working-capital module
The contribution view answers the margin question. It does not answer the cash question. For that, a separate working-capital module is required, and we recommend giving it the same visual prominence as the revenue build.
The module should carry, at minimum, the following inputs.
- Component lead times by category, with explicit safety-stock policy. A 24-week lead time on a critical semiconductor is not an inventory days number; it is a strategic exposure that affects fundraising sizing.
- Supplier payment terms by tier, with concentration analysis. A business carrying 60 per cent of its component spend with a single supplier on 30-day terms is a different credit risk from one with three suppliers on 60-day terms.
- Customer-financing structure. If the business offers a leasing option, the receivable profile and the cost of carrying it must sit in the model. If it relies on third-party financing, the discount or recourse structure should be priced.
- Returns, RMA and warranty reserves. Hardware businesses without an explicit reserve line tend to discover the cost on the cash flow statement two years after launch.
- Foreign-currency exposure on the supplier base, hedged or unhedged, with the natural hedge from any non-sterling revenue netted.
Each input should drive both a balance sheet line and a stress-scenario layer. The committee should be able to ask, in one click, what a 20 per cent extension of average component lead times does to peak working-capital need over the next 18 months. If that question cannot be answered cleanly, the module is not built.
What changes for the committee conversation
Three things tend to happen when a subscription-hardware business is presented through this lens rather than through one of the inherited templates.
First, the headline question shifts from "what is gross margin" to "what is relationship-level contribution and how stable is it across cohorts". That is a more useful question and one the management team is better placed to answer with evidence.
Second, the fundraising sizing conversation becomes specific. Capital requirement is not a single number; it is the sum of operating burn, working-capital build under a defined growth scenario, and a buffer sized to the lead-time and supplier-concentration exposures the working-capital module surfaces. The committee can see why the ask is what it is.
Third, the sensitivity work becomes meaningful. Varying churn alone (the SaaS reflex) misses the point: in a hardware-software business, churn interacts with installed-base replacement cycles, warranty obligations and component sourcing decisions made 18 months earlier. The model needs to express that interaction explicitly. (We address the wider problem of single-variable sensitivity work in a separate note.)
Building the structure
A workable implementation has three layers. The first is a customer or cohort engine that produces relationship-level cash flows for a representative unit and aggregates to the business level under a defined growth path. The second is a contribution P&L that maps those flows to the statutory presentation without losing the underlying structure: the committee sees both views and can move between them. The third is the working-capital module described above, driving the balance sheet and the cash flow statement and linked back to the operating assumptions so that stress scenarios propagate consistently.
None of this is exotic. The discipline is in refusing to import a template designed for a different business. A subscription-hardware company is neither a SaaS company with a manufacturing problem nor a hardware company with a recurring revenue bolt-on. It is a distinct structure, and the model should look like one.
The committee question worth answering is not whether the gross margin is healthy. It is whether the relationship-level contribution, net of working-capital absorption, supports the capital plan under defined stress.
That is a harder question to build for. It is the right one to put in front of a board.