The Hardware Path

Phase 4: Prototype

Building a few units that work like the product and look like the product.

On this page
  1. What the phase involves
  2. Kinds of prototype
  3. Why prototype parts cost so much
  4. What a prototype does not prove
  5. What comes out
  6. Typical cost and duration
  7. What to ask a firm about this phase

The Prototype phase produces the first version of the complete product: one or a few units that work the way the product is meant to work and look close to the way it is meant to look. The question it answers is does the whole product work together, and does it look and feel right?

This is the phase most people have in mind when they say "I need a prototype." It is also the first phase where the spending becomes serious, which is why the three phases before it exist.

Rough guess Assumed Known Possible cost 1ProjectInitiation 2Feasibility 3Proof ofConcept 4Prototype 5Preproduction 6Production
Where this phase sits. The highlighted marker is this phase. Everything to its left has been answered; everything to its right is still a range.

What the phase involves

  • Industrial design. The product's form, surfaces, colors, how it feels in the hand, how it communicates what it does. This is design work in the everyday sense of the word, and it is a separate skill from engineering.
  • Mechanical engineering. Every part designed in CAD: housings, mechanisms, fasteners, seals, how it all assembles. Materials chosen. Tolerances set.
  • Electrical engineering and firmware, if the product has electronics. A custom circuit board replaces the development board from Proof of Concept. Firmware is written for the real hardware.
  • Parts procurement. Custom parts made one or a few at a time: 3D printing, CNC machining, cast urethane, hand-finished. Standard parts bought at small quantities.
  • Assembly and test. Units built by hand, tested against the requirements, taken apart, revised, rebuilt. Several rounds are normal.

The result is a prototype that can be handed to a user, shown to an investor, photographed, and tested in the real environment.

Kinds of prototype

The word covers several different things, and quotes should say which:

  • Works-like. Functions correctly, looks rough. A step up from Proof of Concept.
  • Looks-like. Appearance model with no working internals. Used for photography, user reaction, and packaging design.
  • Works-like, looks-like. Both at once. The usual goal of this phase, and the most expensive of the three.
  • Engineering prototype. Built to test specific things (thermal, drop, water ingress) and not meant to be pretty.

A project might need more than one kind. A looks-like model for a trade show and a works-like unit for testing can be cheaper together than one unit that has to do both.

Why prototype parts cost so much

A custom part in production might cost cents. The same part as a prototype can cost tens of dollars, and the gap surprises almost everyone. The reasons:

  • No tooling. Production parts are made with molds, dies and fixtures that cost thousands up front and then make parts for pennies. Prototypes are made without them, by slower processes.
  • Setup is spread over one part, not ten thousand. A machinist's hour of setup costs the same whether it makes one bracket or a hundred.
  • Hand work. Prototype parts are often finished, fitted and assembled by hand.
  • Small-quantity pricing. Standard components bought in ones cost several times the price in thousands.

The costs page works through a concrete example. The short version: a prototype's unit cost says almost nothing about the production unit cost, and the two should always be two separate lines in a budget.

Key takeaway A prototype proves the product can exist. It does not prove it can be made at a price, in quantity, by someone else. That is the next phase's job.

What a prototype does not prove

A prototype that works on a bench has not shown that:

  • Ten thousand of it can be made with consistent quality.
  • The parts can be sourced from a supply chain rather than a machine shop.
  • The design survives the tolerances of a real factory.
  • It passes the certifications it needs.
  • The production unit cost is anywhere near the target.

Each of these is a Preproduction question. Treating a prototype as a finished product, and ordering a production run from its files, is one of the more expensive mistakes in hardware.

What comes out

  • The prototype units.
  • Complete CAD, drawings, circuit files and firmware. Who owns these should have been settled in the contract before this phase began; see IP and ownership in the guides.
  • Test results against the requirements, with each requirement marked pass, fail or not yet tested.
  • A revised requirements document, now mostly "known."
  • A gate decision: proceed to Preproduction, iterate, or stop. Stopping here with a working prototype is a legitimate end point for a licensing strategy or a fundraising round.

Typical cost and duration

This is where labor hours climb, and labor is most of the cost: industrial designers, mechanical and electrical engineers, firmware developers, technicians. Parts are a smaller share than most people expect. Phases this uncertain are usually billed hourly against an estimate, with the estimate revised at each round.

Common mistake Adding features during the prototype phase. Each new "shall" restarts part of the design. A feature list that was frozen at the end of Feasibility and grows during Prototype is the most common reason a prototype budget doubles.

What to ask a firm about this phase

  • "Which kind of prototype is this quote for?"
  • "How many build rounds does the estimate assume, and what happens if more are needed?"
  • "What will I receive at the end: units, files, test results, all three?"
  • "Who owns the design files, and when?"

Try this (15 minutes) List what the prototype needs to do for the business: convince an investor, test with users, photograph for a campaign, pass a specific test. Then decide which kind of prototype each one needs. The list often shows that two cheaper builds beat one expensive one.

Wayne Weber

Mechanical engineer, MBA. Product development and prototyping.

Mechanical engineer with an MBA, working in product development and prototyping. Questions about anything on this page are welcome by email. Answers are informal, come as time allows, and nothing is for sale.