Skip to main content
IE
Innovate / Engineer
Start Free Sprint
WIKI · STAGE 05 · DEFINE

· Specification

ACTIVITY 05.10.02 · 7 MIN READ

The specification, written.

Also called:  Product specification · Requirements spec · PDS · Engineering requirements

Turning the brief into testable requirements, prioritised Must/Should/Could, with figures, interfaces and standards, so design and manufacture have one unambiguous target.

— TL;DR

A specification is the brief made testable. Every line carries a number, a tolerance or a named standard, and a priority. Prose says “warm and quiet”; a spec says “26°C ±0.5°C, under 30W”. Get this wrong and every later stage inherits the ambiguity.

• • •

What a specification is

A specification is the document that turns “what we want” into “what it must measurably do”. The brief tells you the intent; the spec pins down the figures, tolerances, interfaces and standards that intent commits you to. The test is simple: could a supplier you have never met build to it, and could a tester decide pass or fail without phoning you? If not, it is still a brief wearing a spec’s name.

Every requirement gets a priority, using the M/S/C convention. Must means the product fails without it. Should means painful to lose but survivable. Could means nice if it is cheap. And a fourth column most people skip: Won’t, the things you are deliberately not building, written down so nobody quietly adds them back at week six.

The discipline that separates a spec from a wish list is the acceptance test. For every requirement that matters, write the measurement that proves it. “Holds temperature” is not testable. “Holds 26°C ±0.5°C across a 12-hour run, logged at one-minute intervals, in a 16°C room” is. If you cannot describe the test, the requirement is not finished.

A worked specification canvas

The clearest way to show a spec is to fill it in. Here is the proofing box’s, the one we ran through our pilot, prioritised and made testable rather than left as adjectives.

Specification · the proofing box
MustHold 26°C ±0.5°C for 12h; draw under 30W; UKCA marked to BS EN 61010 electrical safety; fit two 1kg dough balls; mains powered; no app.
ShouldCeramic shell with wood detail; rotary encoder plus OLED control; quiet enough to leave on a worktop overnight without notice.
CouldA second preset temperature, if it adds no meaningful cost and does not complicate the single-dial interface.
Won’tWi-Fi, an app or any account. Out of scope by decision, not omission, to protect the no-app price band and the build budget.
Acceptance test12-hour run in a 16°C room, temperature logged each minute: every reading inside 25.5–26.5°C, mean draw under 30W, plus a passing BS EN 61010 safety report.

Notice each line answers two questions: how good is good enough, and how would we prove it? A bill of materials of £38–55 against a £149 retail price set the cost ceiling every Must had to respect. The figures, not the adjectives, are what the Stoke-on-Trent ceramic supplier and the Manchester PCB shop will actually quote against.

How it fits the bigger picture

The same requirement, written two ways. The left version reads well in a pitch and resolves nothing on a factory floor. The right version tells a supplier exactly what to build and a tester exactly what to check.

✕  Aspirational prose spec
  • “Keeps dough at the perfect warmth.”
  • “Energy efficient and safe.”
  • “Premium, quiet and beautiful.”
  • “Roomy enough for a proper bake.”
✓  Testable requirements
  • Must: 26°C ±0.5°C held for 12h, logged.
  • Must: under 30W; UKCA to BS EN 61010.
  • Should: ceramic and wood; rotary plus OLED.
  • Must: fits two 1kg dough balls.

Each prose line is unfalsifiable: any product can claim it, so none of them settles a quote or a test. Each testable line carries a number, a tolerance or a named standard, and a priority that says how hard to fight for it.

Specification is activity 05.10.02 in the framework, the second activity of Stage 05 Define. It builds on the design brief that opens Define, and feeds straight into the moodboard (05.10.03), where the Should-level look and feel gets pinned down before any of it reaches CAD.

01 02 03 04 05 06 07 08 09 10 Idea Discover Innovate Evaluate Define Design Engineer Develop Manufacture Deliver YOU ARE HERE

What it can do

It gives design and manufacture one unambiguous target. A supplier can quote against it, an engineer can design to it, and a tester can sign it off against the acceptance criteria. The M/S/C priorities also tell everyone what to trade away first when, inevitably, cost or time gets tight.

What it can’t do

It can’t decide whether the requirements are the right ones; that judgement comes from the Discover and Evaluate work behind it. And a spec written too tight, too early, freezes choices you have not earned yet. Pin the Musts hard, hold the Shoulds and Coulds loosely, and let later stages tighten them.

See the full 10-stage process →

Try it yourself

Take your design brief. Rewrite every requirement so it carries a number, a tolerance or a named standard, then tag each one Must, Should, Could or Won’t. For every Must, write the test that proves it passes. Anything you cannot turn into a test is still a wish, not a requirement, so either make it measurable or move it down a priority.

Want a guided first pass? Start the Free Sprint → and the GPT will help you turn the brief into measurable requirements.

Your specification checklist

Project notes: the half-degree that set the budget

  From the notebook · optional reading

Writing the proofing box spec with Dan and Anna Hartley in Stockport, and the one tolerance figure that quietly decided what the whole product would cost.

3 min read · click to open

Dan arrived with a one-line brief: “a beautiful box that keeps sourdough warm overnight, no faff.” Lovely sentence. Useless to a factory. I asked the question that turns a brief into a spec: “How warm, how steady, and how would we prove it held?”

The number that set everything else

Anna knew her dough: 26°C, and the tighter she held it the more predictable the prove. We pushed on the tolerance, because tolerance is where the cost lives. ±2°C is a cheap thermostat. ±0.5°C needs a proper sensor, better control and a well-insulated ceramic shell. We wrote ±0.5°C as a Must, and accepted that this one figure pulled the bill of materials up toward the £38–55 band the £149 price could carry, and no higher.

Sorting the rest into Must, Should, Could, Won’t

  • Must. 26°C ±0.5°C for 12h, under 30W, UKCA to BS EN 61010, two 1kg dough balls, mains, no app.
  • Should. The ceramic-and-wood shell and the rotary-plus-OLED dial. Lose them and you still have a working product, just a cheaper-feeling one.
  • Could. A second preset temperature. We left it as a Could and, sure enough, it was the first thing cut when the interface got busy.
  • Won’t. Wi-Fi and an app, written down explicitly so the idea couldn’t creep back in.

The line that earned its keep was the acceptance test. I wrote: 12-hour run in a 16°C room, logged every minute, every reading inside 25.5 to 26.5°C. Months later, when the first ceramic prototype drifted to 27.1°C on hour nine, there was no argument about whether it had passed. The spec had already decided. We sent it back to the Stoke-on-Trent supplier with one figure and one chart, not a paragraph of opinions.

That is the whole point of writing it down properly. The half-degree we argued over in week one settled a dozen disputes we never had to have later.

— Define stage, project notes, 2026

— Next in Define → Moodboard