MVPs for physical products
Software founders treat the MVP as obvious. Hardware founders treat it as impossible. Both are wrong in different directions.
The trap with physical products is the assumption that minimum means finished. It does not. It means small, ugly, and in front of a real person within four weeks. Anything else is a side quest.
Don’t fabricate. Assemble.
The default move when you sketch a physical product is to design custom parts. Don’t. Custom anything pushes the first ship date out by months and tens of thousands of euros, and you don’t yet know if anyone wants it.
Your factory for the first ten units is Aliexpress, Mouser, Amazon, and whatever you can 3D-print overnight. Buy modules, glue them together, write the firmware that makes them behave. The bill of materials should read like a Saturday-afternoon shopping list, not an engineering BOM.
The ugliness is the point. Off-the-shelf parts force you to validate the function, not the form. Form is a problem for unit eleven.
Four weeks. Calendar discipline.
Pick a customer. Pick a date four weeks out. Whatever you ship on that date is the MVP, by definition. Then do it again, four weeks later.
The trap inside hardware is the multi-month roadmap that quietly slides into a multi-year roadmap. Four-week loops break that. They force ruthless cutting — you can build the temperature sensor or the display, not both. The customer reaction tells you which one mattered.
The calendar beats the spec sheet. Every time.
One named buyer, not “users”
“Users” is the most expensive word in product. It lets you imagine a patient, generic audience that doesn’t exist. Real validation comes from one named person who hates the problem so much they will tolerate your duct tape.
Find that person. Hand them the device. Watch their face. Write down what they do, not what they say. The face is the data.
If you can’t name your first buyer, you don’t have an MVP — you have a prototype with marketing aspirations.
Listening is not surveys
Customers don’t know what they want until they hold it. So don’t ask them what they want. Ship them something, watch them use it, and adjust.
The signals that matter, in order:
- What they pay — money is the only honest signal. Discounts and pilots don’t count. A signed PO does.
- What they do unprompted — second time they pick it up without you asking, you’ve got something.
- What they complain about — complaints map directly to the next four-week loop.
- What they say in a survey — least reliable, ignore unless triangulating.
Build for one. Charge them money. Iterate against what they actually do. Then earn the right to talk about “users” later.
The compounding part
Each four-week loop teaches you something you couldn’t have learned from the desk. The teaching is the asset. By the fourth or fifth loop, you know more about this specific customer’s actual problem than anyone else in the market — including the customer.
That knowledge is the moat. It is also the only thing you can’t fake by hiring an industrial designer or buying a CNC.
Off the shelf. Four weeks. One buyer. What they pay. Repeat.
The physical MVP is not slower than the software MVP. It is the same loop, with shipping costs.