Startups are Experiments
New ventures operate in conditions where little is known and much is guessed. Treating work as a series of small, honest experiments turns guesswork into learning before money and time are burned on features no one wants. A detailed plan can feel safe, yet it is built on assumptions about problems, customers and channels. Progress comes from exposing those assumptions to reality as early as possible and adapting based on what you learn.
What Makes A Good Experiment
State a clear hypothesis that can be proven wrong. Specify the customer segment, the problem you believe they face, the behaviour you expect to see, and the measure that will confirm or reject it within a set time. Keep it small, fast and cheap. Focus on the riskiest assumption first, which is usually about value: do people care enough to take action? Only after value is evidenced should you worry about scaling or automation. Choose a single success metric and a threshold that counts as a pass, such as sign-up rate, activation rate or weekly retention. Decide these before you run the test to avoid wishful thinking.
Minimum Viable Product
Build the smallest thing that lets a real customer show interest or use the core value. This might be a landing page with a clear promise and a waiting list, a concierge service where you do the work manually behind the scenes, a simple prototype video, or a price test that asks for a card but does not charge until you are ready. For example, a meal kit idea can be tested with a one-page site, modest ads targeting busy professionals, and a form that captures postcode and delivery preferences. If enough visitors sign up within a week at the proposed price, you have evidence that the problem and offer matter. Follow with a short, manual pilot to observe real usage and reasons for drop-off.
Measure, Learn, Decide
Collect only the data needed to judge the hypothesis. Prefer cohort measures and rates over vanity totals, and pair numbers with short interviews to understand the “why”. When results meet or beat the threshold, double down on the next riskiest assumption, such as willingness to pay or referral. When they fall short, change course deliberately: adjust the segment, refine the problem, or reshape the offer. Keep ethics front and centre; be clear about waitlists, avoid surprise charges and close the loop with those who took part. Work in tight loops: define, test, learn, decide, repeat. Over time, a chain of small, fast experiments replaces rigid planning with evidence-led progress.