The Model Is Part of the Design, Not a Postscript

Founders usually pick a business model after the product is engineered, and by then most of the options are gone. A device with no connectivity cannot support a subscription. A cartridge with no authentication cannot defend a consumables stream. A product built without a serial number, a service port, or field-replaceable modules cannot be leased and refurbished. Every one of those is a decision made during architecture, often by an engineer who was never told what the revenue model was supposed to be.

So the sequence has to run the other way. Pick a candidate model early, list what it demands from the hardware, and price those demands into the development budget. Changing your mind later means a respin.

Five Models and What Each One Costs You to Build

One-time sale. You sell the unit, the transaction ends. Simplest to build and to explain, and it puts cash in on day one. The problem is that every dollar of next year's revenue requires a new customer, and for a durable product with a seven-year life you are selling into a market that saturates. It also puts brutal pressure on gross margin, since the whole business runs on the spread between landed cost and price. If this is the model, spend the effort on value engineering early, because a dollar off the bill of materials is pure profit forever.

Product plus consumable. The device is priced near cost, sometimes below it, and the money is in the refills, cartridges, filters, pods, or test strips. Very strong when the consumable is genuinely necessary and reasonably hard to substitute. It demands real engineering: a keyed interface, an authentication scheme, quality control on the consumable supply chain, and enough regulatory care that a third party cannot ship a cheaper compatible next quarter. The economics and the traps are worked through in the razor-and-blade model.

Subscription for a connected service. The hardware sells at a normal margin and a monthly fee buys cloud storage, analytics, alerting, or app features. This is the model most connected-product founders default to, and it is the one most often modeled wrong, because they forget that serving the device costs money whether or not the customer pays. Run the per-device monthly infrastructure number before setting a price; the drivers are in IoT cloud costs. A subscription also creates an obligation: if the service dies, the product bricks, and customers know it.

Product as a service. The customer never owns the unit. You lease it, maintain it, and take it back. Common in commercial equipment, medical capital gear, and industrial monitoring. It converts a capital purchase into an operating expense for the buyer, which shortens sales cycles dramatically in B2B, and it turns your balance sheet into a financing problem, because you fund the hardware and recover it over 24 to 48 months. The engineering requirements, remote diagnostics, tamper detection, refurbishment design, are covered in hardware as a service.

Licensing the technology. Somebody else manufactures and sells; you collect a royalty. Lowest capital requirement, lowest control, lowest ceiling. Typical royalty rates on consumer hardware land between 3 and 7 percent of wholesale, which sounds thin until you compare it against the capital needed to do it yourself. The full comparison is in licensing versus manufacturing.

Three Numbers That Actually Decide

Everything above collapses into three quantities, and if you cannot estimate them, you are not ready to choose.

  • Purchase frequency. How often does the same customer buy again? A product replaced every ten years cannot run on one-time sales alone unless the unit price is high or the market is huge.
  • Cost to acquire a customer. Digital acquisition for a physical product commonly runs $40 to $200 per customer, and higher in competitive consumer categories. If acquisition costs $120 and the one-time gross profit is $90, the model is broken no matter how good the product is.
  • Gross margin per unit and per month. Not the factory quote, the landed cost with freight, duty, returns, warranty reserve, and channel margin. Getting from that number to a shelf price is its own exercise, laid out in how to price a product.

The model that wins is usually the one where lifetime value clears acquisition cost by at least three times, with payback inside twelve months. Hybrid structures are normal: sell the hardware at a healthy margin, attach an optional service tier, and add a paid extended warranty for buyers who want it. Roughly a third of buyers take a well-priced service plan, and it carries very high margin.

Test the Model Before You Commit the Tooling

You can validate a pricing model long before production. Put the actual price and the actual terms on a landing page and see whether people convert. Sell a small batch of pilot units by hand at the intended price and watch how many attach the subscription. Interview ten buyers in your target segment about how they budget, because in B2B the difference between a capital purchase and a monthly fee determines who has to approve it.

What you cannot do is survey people about willingness to pay and believe the answers. Only a transaction is evidence. Broader channel and go-to-market context sits on the business development hub.

Work the Model Into the Engineering

Projects House builds products with the revenue model in the requirements from the first review, whether that means a keyed consumable interface, a device identity scheme for subscription entitlement, or a design that survives refurbishment. Tell us what you are building and how you plan to make money on it through our contact form.