Experiments
Set a stopping rule for a small experiment
An optional worksheet for writing the limits, early signals, and change triggers that tell you when a small experiment should continue, shift, or end.
Many small experiments never really end. Sessions get skipped, the tracking sheet stops updating, and months later nobody can say whether the idea failed or simply lost attention. A stopping rule written at the start gives the experiment a defined edge.
This guide offers an optional worksheet for that rule. It is a planning aid, not a tested framework. Use the parts that fit, skip the rest, and treat each condition you write as your own judgment, made in advance.
Decide whether the experiment needs a rule
A stopping rule is worth writing when the experiment costs something you will notice: weekly hours, money, a promise to other people, or energy you need elsewhere. A five-minute trial you can drop without consequence probably does not need one.
If you write one, do it before the first run, and keep it short enough to reread at every check-in. The worksheet has three working parts: ceilings, early signals, and a change trigger. You continue when none of them fires.
Write the ceilings first
A ceiling is the most you will spend before the experiment has to stop as planned. Use a unit you can count, such as preparation hours, money, calendar weeks, or sessions run. Say how often the ceiling may be exceeded. One heavy week can be noise. Two in a row is a pattern you agreed to act on.
Ceilings ignore whether results look good, and that is deliberate. A promising experiment that takes twice the planned time is no longer the experiment you agreed to run.
Name the signals that end it early, and the change trigger
An early signal ends the experiment whatever the budget says. Examples include no participants twice in a row, someone being harmed or seriously inconvenienced, or a new fact that removes the reason for the test. Write each signal as something you could observe and record.
A change trigger means the idea still seems sound but its current form is not working. Write what you would alter and what stays fixed. If the change would alter the question itself, start a new experiment with a new rule instead of editing this one.
| Part | What to write | When to check |
|---|---|---|
| Ceiling | A countable limit and how often it may be exceeded | Every check-in |
| Early signal | An observation that ends the experiment regardless of budget | As soon as it occurs |
| Change trigger | What is not working, what you would alter, what stays fixed | Every check-in |
| Closing note | Which condition fired, the date, and what was learned | When the experiment ends |
Worked example: evening study sessions
Ines, a fictional tutor, plans four weekly 30-minute evening review sessions. Before week one, Ines writes the rule. The ceiling: stop if preparation exceeds 2 hours in two consecutive weeks. The early signal: stop after zero sign-ups in two consecutive weeks. The change trigger: if more than half of attendees leave before the halfway point, switch to a shorter question-and-answer format.
Week one has 2 sign-ups and 2.5 hours of preparation. Week two has 4 sign-ups, nobody leaves early, and preparation again takes 2.5 hours. Attendance is rising, but the ceiling has been exceeded two weeks running, so the experiment stops as written.
The closing note records that the ceiling fired at week two, that attendance rose from 2 to 4, and that most preparation went into writing practice sheets from scratch. A new experiment with its own rule could test whether reusing sheets keeps preparation under 2 hours. The original rule stays unedited.
Limitations
The worksheet cannot tell you whether a threshold is sensible. A careless rule can stop a good idea early or keep a weak one running. Small experiments produce small numbers, and a rise from 2 to 4 attendees is too little to read as a trend. If something outside the experiment forces an end, write that in the closing note rather than claiming a condition fired.