Make the start smaller
For the report problem, write three rough bullets before editing anything. Test whether an explicit first output makes beginning feasible.
Practical guide / updated October 24, 2026
Turn an overthought plan into one weekly behavior experiment.
A useful experiment answers a smaller question than "How do I change my life?" It asks whether one particular action fits one recurring situation. This guide covers the decision before the week: choosing what to test, containing further planning, and knowing what the result can tell you.
What is a weekly behavior change experiment?
It is a seven-day observation window with one explanation to test, one chosen method, a clear cue, and an observable action. You decide what counts before starting. At the end, you use the record to keep, reduce, switch, or replace the method. Seven days is the appointment for a decision; it is not a deadline for becoming a different person.
1. Define the uncertainty that planning has not resolved
Start with a sentence describing a repeated event. "I open messages before beginning my report" contains something observable. "I am an unfocused person" does not tell you which action or context to change. Separate the desired outcome, such as submitting a report, from the behavior you can test, such as writing its first three bullet points before opening messages.
Then write a tentative explanation: "I may delay because the first step is unclear." That explanation might be wrong. The experiment should produce information that could challenge it. If the first step is made explicit and starting remains difficult, you have a reason to investigate a different obstacle. Do not describe that tentative explanation as a diagnosis.
2. Put a boundary around choosing
When every plan creates another plan, establish four limits: one problem, three candidate methods, one signal, and one review date. Three is the choice constraint used by ChangeLoop 7, not a scientifically proven ideal number. Its practical purpose is to let you compare alternatives without endlessly requesting another list.
Prepare the materials
Leave the relevant document and notes ready. Test whether searching for materials was the immediate barrier.
Change the cue
Start after a dependable event rather than at an unreliable clock time. Test whether timing was the mismatch.
These are fictional planning examples, not customer results. Select one approach for this cycle. Combining all three would make an apparent improvement harder to interpret. Record the unchosen options in a parking list; a new idea can wait for the planned review unless circumstances make the current action unsuitable.
3. Set a signal you cannot quietly redefine
For this fictional report experiment, the minimum is three bullets written before messages, within five minutes of opening the work document. The signal is whether those three bullets existed before messages were opened. Writing a complete paragraph is optional. Thinking about the report does not count. Opening the document without writing does not count either.
The five-minute limit is an example chosen to make the test affordable, not a research-backed universal dose. Decide whether your cue can occur on all seven days. A work-specific cue may never occur on a weekend. In a manual worksheet, identify those absent opportunities explicitly; do not present them as successful actions. In the product, keep Done and Not done truthful and explain a missing opportunity in the weekly review.
4. Keep planning from consuming the trial
Choose a start date and place the review in your calendar before collecting more advice. You can set a personal planning budget, such as 15 minutes, to finish the first draft. That is a suggested constraint, not a tested treatment. When you encounter a new technique on day three, ask whether it answers an immediate safety or feasibility problem. If it only looks more exciting, leave it for the next review.
A stable plan does not require forced persistence. Stop an action that causes harm, becomes inappropriate, or depends on circumstances that no longer exist. Note what changed. If you change the cue or minimum during the week, record the old and new versions separately. A week containing two methods cannot cleanly answer a question about just one of them.
5. Read the result as information about fit
Imagine the fictional log contains four completed actions, two missed actions after unexpected meetings, and one day without the work cue. That is not a 100% success rate because the unavailable day was ignored. It is a description of four completions across seven calendar days, including one absent opportunity. You can now ask whether moving the cue would address the specific scheduling problem.
Even seven completions cannot establish why the method worked. A lighter workload, novelty, or an unusually quiet week could contribute. A repeated cycle under ordinary conditions can provide additional observations, but a personal log is not a controlled trial. Use it to make a proportionate next decision, not to claim a permanent habit or prove a theory about everyone.
What research supports, and where this guide adds its own choices
Lally and colleagues' original habit-formation study followed 96 volunteers over 12 weeks. Its modelled times to near-maximum automaticity varied widely. That supports rejecting a universal seven-day habit deadline. Source accessed October 24, 2026.
Harkin and colleagues' progress-monitoring meta-analysis included 138 studies and 19,951 participants, finding an average benefit for goal attainment. It did not test this website or establish our exact weekly schedule. Source accessed October 24, 2026.
The three-option boundary, report scenario, and planning budget above are practical design examples. For a blank form, use the behavior experiment worksheet. For interpreting an already completed week, use the weekly review template. The distinction keeps planning and evaluation from becoming the same unfinished task.