Many business decisions are based on opinions: what the founder believes customers want, what competitors seem to be doing or what feels right. Opinions are useful starting points, but they can be wrong. Evidence reduces uncertainty by showing what actually happens.
Small tests allow businesses to gather evidence quickly, cheaply and with limited risk.
Identify the key uncertainty
Before testing, identify the assumption that matters most. For example, will customers pay a higher price? Will a new service attract enough demand? Will a process change reduce errors?
Focus testing on the uncertainty most likely to affect the decision.
Design a small test
A good test is small enough to run quickly but realistic enough to provide meaningful evidence. Examples include a pilot project, a limited-time offer, a trial with a small group of customers, a short advertising test or a process trial for one team.
Define success before testing
Decide in advance what result would support the decision and what result would suggest changing course. Clear success criteria prevent interpreting results based on hope.
Use existing data
Before creating a new test, review existing evidence: sales data, customer feedback, website analytics, previous projects or industry information. Existing data may answer the question without additional effort.
Talk to customers
Customer conversations can reveal needs, objections and preferences. Ask about past behaviour rather than hypothetical intentions. People are often better at describing what they have done than predicting what they will do.
People are often better at describing what they have done than predicting what they will do.
Measure carefully
Track relevant outcomes during the test: sales, sign-ups, costs, time saved, errors or satisfaction. Keep measurement simple and consistent.
Learn from unexpected results
Tests do not always confirm assumptions. Unexpected results provide valuable lessons and may reveal better opportunities.
Scale gradually
If a test succeeds, expand gradually. Larger rollouts may reveal new challenges that did not appear in the initial test.
A practical tool: the test card
Before each test, complete a card with five lines: the assumption being tested, the test itself, how long it will run, the measure of success and the decision you will make depending on the result. After the test, add the actual result and what you learned. A collection of test cards becomes a valuable record of evidence.
Questions before testing
Ask: What is the riskiest assumption in this decision? What is the cheapest way to learn whether it is true? How long will the test need to run for a meaningful result? What will we do if the result is ambiguous? Is the test realistic enough to reflect how customers or the team will really behave?
Common mistakes
Businesses often run tests without clear success criteria, rely on opinions instead of evidence, test too many variables at once or scale too quickly after limited success. Others ask customers hypothetical questions and treat the answers as reliable predictions. Clear, small and well-measured tests produce more useful evidence.
An example in practice
A small bakery considered adding a subscription service for weekly bread deliveries. Instead of investing in new equipment and marketing immediately, it offered the service to fifty existing customers for one month. Thirty-two signed up, and most renewed. Using this evidence, the bakery expanded the service gradually and adjusted delivery routes based on what it learned during the trial.
When testing is not possible
Some decisions cannot be tested in advance, such as signing a major contract or making a one-time investment. In these cases, look for indirect evidence: experiences of similar businesses, expert advice, historical data and small related experiments. Even partial evidence improves a decision compared with relying on opinion alone.