Experiment design ·

Before running a test, name the decision it can change

A technically good test can still leave the programme exactly where it started.

Joseph Wright of Derby's alchemist kneels before a glowing flask while two assistants work in shadow.
Joseph Wright of Derby · The Alchymist, in Search of the Philosopher's Stone, Discovers Phosphorus · 1771 · Derby Museum and Art Gallery · public domain · image provenance

Industrial programmes can produce excellent microscopy, simulations, surveys, and reports without resolving the uncertainty that blocks investment. The problem is often set before the test starts: nobody has agreed what any possible result will permit.

Write the decision sentence first

A useful test brief can fit on one page. It should answer five questions:

  1. Which decision is blocked?
  2. Which uncertainty blocks it?
  3. What is the smallest observation that addresses that uncertainty?
  4. Which threshold separates the possible outcomes?
  5. What action follows from each outcome?

The final question gives the evidence authority. A positive result might release a pilot batch. A negative result might stop one process route. An inconclusive result might justify one repeat under a specified condition.

A quick failure check

Ask what happens after a positive, negative, or ambiguous result. If every branch leads to another test, the work may create knowledge, but it is not yet governing a development commitment.

Stage-gate methods make the consequence explicit with choices such as proceed, hold, repeat, or stop. Their value here is practical: the team agrees in advance how evidence changes action.

Exploratory research deserves a different standard. Discovery work can be valuable before its application is known. Once a programme claims to reduce development or market risk, however, its experiments should identify the uncertainty, threshold, and consequence they address.

Sources