Every stage of building a hardware product, explained plainly. What to do, what goes wrong, what it costs when it does, and what tools and resources actually help — written by someone who has done it.
Find the stage your product is currently in and read that section first. Each stage covers what it is, the most expensive mistakes teams make there, what you need to have ready, and which tools or resources apply. The mistakes aren't hypothetical — they're drawn from real programs, with real costs attached. If you want the full picture, the book goes deeper on every stage.
The PRD is the contract between your product vision and your engineering team. It defines what the product must do, what it must not do, and the constraints it must operate within — regulatory requirements, environmental ratings, target cost, weight, dimensions, battery life. Every engineering decision that follows will reference this document. If it is vague, every subsequent decision will be vague too.
The prototype stage is about proving the concept works at a bench level — not about proving it will manufacture. This is the cheapest point in the programme to make mistakes. A prototype that is wrong costs thousands to fix. The same mistake found in PVT costs hundreds of thousands. The purpose of prototyping is to find as many failure modes as possible before money is committed to tooling or processes.
EVT is the first build using production-intent tooling and processes. The goal is to validate that the engineering design works as specified — not to prove it is ready to manufacture. Units are typically built by hand or in very small quantities (10–100). Every discipline — mechanical, electrical, RF, software, battery — runs their critical validation tests. Issues found in EVT are expected. Issues that carry over to DVT without resolution become expensive.
DVT validates that the design is complete and ready to manufacture at volume. Units are built closer to production conditions — typically 50–200 units — and every qualification test runs in full. Regulatory submissions are prepared. The CM is being qualified. The supply chain is being stress-tested. DVT is the last stage where design changes are manageable. Changes in PVT are painful. Changes after mass production starts are catastrophic.
PVT is the trial production run — typically 200–1,000 units built on the actual production line, by production operators, with production tooling, to prove the line can produce the product at acceptable yield. The engineering focus shifts from design to process. Yield problems, assembly sequence issues, and EOL station failures are the failure modes that surface here. A PVT gate pass authorises mass production.
Selecting a contract manufacturer is one of the most consequential decisions in a hardware programme. A good CM is a partner who catches problems, suggests improvements, and protects your quality. A bad CM produces units that look fine at inspection and fail in the field. The decision is almost impossible to reverse without significant cost and delay. Most founders make it too late, too quickly, and based on price alone.
Mass production is when the programme stops being an engineering project and becomes an operational discipline. The design is fixed. The process is fixed. The goal is consistency — producing units to spec, at yield, on schedule, every day. Engineering's role shifts to monitoring, root cause analysis of yield excursions, and managing engineering changes. Surprises at this stage are expensive by definition.
Sustaining engineering is the ongoing work of keeping a product in production — managing component obsolescence, cost reduction initiatives, quality improvements, and regulatory re-certification when standards change. End of life is the planned, controlled wind-down of a product — managing last-time-buy inventory, customer communication, spare parts obligations, and regulatory deregistration where applicable. Both are frequently neglected until they become crises.
Every service and resource mentioned in this guide, in one place.