Hardware and SaaS are not the same beast, and thinking of one as the other with atoms bolted on is the first mistake every founder who moves from software to physical products has to learn out of, and most of them learn it the expensive way. I have shipped over two hundred thousand orders through Murphy Door, across three American plants in Ogden, Lexington, and Rigby. The costs, the timelines, and the failure modes of scaling a hardware company are their own beast. This is what I have actually spent to learn them.
The cost is not the cost of goods
New hardware founders obsess over cost of goods. They should. It is a real number and it matters. But cost of goods is a small fraction of the real cost of running a hardware company at scale. The real costs live in inventory carry, in freight, in returns, in warranty, in the tools that make the plant run, and in the working capital you tie up between the day you buy the raw material and the day the customer's payment clears.
A hardware company that grows fast is a company that borrows a lot of money to buy things it won't sell for months. That is what growth looks like in physical products. Software growth is a capacity conversation. Hardware growth is a capital conversation. Two very different companies, two very different constraints, two very different sets of people who will be helpful.
Most new hardware founders underestimate this by an order of magnitude and end up in a cash squeeze the first time demand really turns on. That squeeze is a working capital problem masquerading as a demand problem, and it fools founders who have never seen it before.
Lead time is a strategy
In software, lead time is one week. In hardware, lead time is anywhere from six weeks to nine months depending on the product and where the components come from. Founders who treat lead time as an inconvenience will always be behind. Founders who treat lead time as a strategy will win.
What that means in practice is that you plan two seasons ahead, not one. You place raw material orders based on the season after next. You forecast SKUs by month and by plant. You get comfortable with the fact that a decision you make today about wood species and hardware alloy won't show up in a customer's home for four months, and if you are wrong, you will still be shipping the wrong SKU in month four.
The founders who scale hardware well are the ones who develop a forecasting muscle their software peers don't need. That muscle is worth more than any marketing tactic, because a hardware company with a good forecast can outrun a hardware company with better product for a long time.
Returns are the real margin killer
In software, a customer who wants a refund clicks a button. In hardware, a customer who wants a refund puts a door back on a truck and sends it across the country to your dock, where you have to inspect it, re-stock it or scrap it, refund the customer, and pay for the freight in both directions. A single return can cost you the margin from three orders.
Reducing returns is the single most valuable cost activity in a hardware company. Not marketing. Not manufacturing efficiency. Returns. Every dollar you spend on better product photography, better fit tools, better install instructions, better customer service before the sale, all of it returns as multiples on the return rate. And every point you shave off the return rate hits the P&L twice, once through fewer refunds and once through less freight and rework.
We measure returns weekly at the SKU level and we investigate every SKU that spikes. Some of the time the fix is a better product page. Some of the time it is a hardware change. The rest of the time it is an installer training issue. All three fixes are cheaper than one return.
Warranty is a forecast, not a promise
Hardware founders write a warranty policy and then move on. Warranty is a forecast about how many parts you will have to replace in year two, three, and five, and how much that will cost you. It is not a policy you can write once and shelve. If you get the forecast wrong, you will be paying warranty out of current cash flow instead of a reserve, and the finance team won't have seen it coming.
The way we handle warranty at Murphy Door is that every quarter we set aside a percentage of current-quarter revenue as a warranty reserve, based on our best rolling estimate of what year-two through year-five failures will cost us. When a real warranty claim comes in, it hits the reserve, not the current P&L. That accounting move sounds boring. It is the single most important accounting decision a hardware company makes.
Companies that don't do this end up with a bad quarter in year three when the year-one products all start failing at the same time, and they think it is a manufacturing problem. The real cause is an accounting problem they created for themselves in year one by not reserving.
The plant is a fixed cost that does not forgive
Software fixed costs are easy to shrink. A hardware plant is not easy to shrink. Once you have signed a lease for square footage and hired a shift, you are carrying that cost every month whether or not the orders are coming in. Plants punish founders who haven't learned to smooth demand.
Smoothing demand in a hardware business means seasonal SKUs, contract manufacturing for adjacent brands, wholesale channels that buy in predictable volumes, and a direct-to-consumer forecast that is honest about the peaks and valleys of the year. Founders who build a plant for the peak week and then run at forty percent capacity for the rest of the year go bankrupt. Founders who build a plant for the average week and figure out how to serve the peak week through smart overtime and inventory pre-builds survive.
This is why the three-plant setup at Murphy Door works. We can flex orders between Ogden and Lexington by geography, and we can pull from Rigby for material to keep both plants fed. A single-plant hardware business is fragile at scale. Multi-plant is the answer, but multi-plant is only affordable once demand justifies it. Getting the timing right on the second plant is the hardest single decision a hardware founder makes.
Hardware teaches you what software can't
Here is the honest reason I keep building hardware companies. Hardware teaches you things software can't. It teaches you about supply chains, about labor, about tools, about waste, about people who work with their hands, about towns that need employers, about parts of the economy that are hidden from most founders who work only in software.
The founders I know who came from software and moved into hardware are, universally, better founders after they have done it. Not because software is easy. Because hardware forces you to see the physical world in a way software does not. You can't un-see it once you have.
The founders I know who came from hardware and moved into software are also better founders after they have done it. The direction does not matter. What matters is that you have worked in both, so that you know what the other side actually looks like.
What I tell founders thinking about hardware
Do not start with a plant. Start with a product people paid you for. Get the first hundred orders out of a garage or a small contract manufacturer. Get the first thousand orders out of a small facility you rent by the month. Only sign for a real plant when the orders exceed what a rented facility can produce, and when you have a real forecast that says the orders will keep coming for the next twenty-four months.
Reserve for warranty from month one. Measure returns weekly. Forecast raw material by season, not by month. Smooth demand across seasonal SKUs and adjacent channels. Hire the plant manager before you hire the marketing lead, because the plant manager is the one who will keep you out of bankruptcy in year three.
Hardware is hard, and ours has shipped over two hundred thousand orders. That number is a count, not a projection. The cost of scaling it is real, and it is not what most founders think it is going in. Now you know.