Most failed health experiments do not fail because the intervention was useless. They fail because nobody can tell what happened.
The baseline was vague. Three variables changed together. Adherence was assumed. The outcome was never defined. The retest happened too early, too late, or not at all. A strong feeling became the final data point.
If you cannot describe the experiment, you cannot trust the conclusion.
Define one decision
Begin with a decision, not a product. "Should I continue this intervention?" is better than "Is this supplement good?" The first question forces a target, a time window, and criteria. The second invites an argument.
A useful protocol record has seven parts:
- Target: the symptom, function, or marker that should change.
- Baseline: the starting value and how it was measured.
- Intervention: the exact behavior, product, drug, dose, timing, and source.
- Adherence: what was actually done, not what was prescribed.
- Confounders: other changes that can move the outcome.
- Review date: when enough time has passed to judge the result.
- Stop rule: the side effect, threshold, or lack of benefit that ends the experiment.
Choose measurements that answer the question
Different arrows need different combinations of objective and subjective measures.
Body may use blood pressure, waist circumference, body weight trends, lipids, glucose markers, strength, or cardiorespiratory performance. Mind may use a validated symptom questionnaire, focus blocks completed, mood ratings, or clinician assessment. Recovery may use sleep timing, estimated duration, daytime sleepiness, training readiness, and symptoms of a sleep disorder. Vitality may use hormone measurements, sexual-function symptoms, fertility measures, and energy ratings.
No single number owns the conclusion. Laboratory precision does not make the wrong marker useful. A subjective measure is not worthless because it is subjective. It needs consistent collection and a clear relationship to the goal.
Change one major variable
Perfect isolation is rare outside a laboratory, but avoid creating preventable ambiguity. If you start a medication, change diet, add three supplements, and begin a new training program together, improvement will be hard to attribute and side effects will be hard to trace.
Single-person experiments, sometimes called N-of-1 trials, formalize repeated observation within one person. A home protocol does not become a clinical trial because it has a spreadsheet, but the same logic helps: make the intervention explicit, keep measurement consistent, and compare defined periods.
Adherence is a result
An intervention that works only when followed perfectly may fail in real life. Record missed doses, skipped sessions, travel, illness, alcohol, major sleep disruption, and changes in routine. If adherence is low, do not label the biology ineffective. Decide whether the protocol is too burdensome or whether it needs a fairer test.
Cost and friction belong in the result. A small benefit from an expensive, unpleasant, indefinite protocol may not be meaningful. Minimum effective dose includes effort.
Retest on purpose
The review interval depends on the intervention and outcome. Blood pressure can move quickly. Hair growth and body composition take longer. Hormone testing needs attention to timing and clinical context. Sleep trends need enough nights to separate a pattern from noise.
Set the interval before the experiment starts. At review, choose one of four actions: continue, adjust, stop, or refer. Avoid the fifth action, which is to keep paying while promising to evaluate later.
Know when measurement is not enough
Self-tracking cannot rule out disease, interpret every abnormal result, or make a risky intervention safe. Severe or rapidly worsening symptoms, red-flag findings, concerning side effects, and decisions involving prescription drugs require qualified clinical care.
Four Arrows is built around a closed loop: baseline, decision, action, measurement, and review. The tools will change. The discipline should not.