One test, one question
A good experiment answers one question: "Will restaurant owners reply to a message about review management?" or "Will five people pay $50 for this template?" If a test tries to answer everything, it answers nothing.
Decide the measure before you start
Write down what success looks like before you launch: "At least 3 replies out of 20 messages" or "2 pre-orders in a week". Deciding afterwards makes it too easy to call any result a success.
Make it small and fast
Most tests should take less than a week and cost almost nothing. Use what you already have: your network, a simple page, a message, a call. Speed matters more than polish at this stage.
- Landing page with a "join the waiting list" button
- Ten direct messages with a specific offer
- A free 20-minute audit offered to five businesses
- A sample delivered to one client at a pilot price
Stop, change or continue
After each test, choose one of three actions. Continue if the measure is met. Change one element (target, message, price) if results are close. Stop if there is clearly no interest. Stopping early is a win: it saves you months.
Example
Two tests in one week
Test 1: 20 messages to gyms offering social posts. 1 reply: stop. Test 2: the same offer to physiotherapists. 4 replies and 2 calls booked: continue with physiotherapists.
Prompt to try
Replace the [brackets] with your own details and review the result before using it.
Help me design a one-week experiment. My idea: [idea]. My target: [who]. Give me: the single question to answer, the success measure, the cheapest way to test it, the steps day by day, and what I will do if the result is good, close or bad.
Exercise
Design one test you can run in five days.
