What does it mean that the experiment review?

Regularly review your experiment results to extract patterns.

Example: You have been running behavioral experiments for eight months. You have forty-three entries in your experiment journal — sleep interventions, work routine modifications, communication experiments, exercise variations, dietary changes. Individually, you have evaluated each one: this worked, that failed, this was ambiguous. But you have never sat down and read them together. When you finally block ninety minutes for an experiment review, patterns emerge that no single entry could have revealed. You notice that every experiment involving morning routines succeeded at a higher rate than evening routines — not because mornings are inherently better, but because your evenings are consistently disrupted by unpredictable family commitments that contaminate your experimental conditions. You notice that experiments involving subtraction — removing a behavior — have a 70% success rate, while experiments involving addition — adding a new behavior — succeed only 35% of the time. You notice that your three most impactful experiments all shared an unexpected feature: they changed your physical environment rather than relying on willpower. None of these insights existed in any individual record. They became visible only through systematic review across records. You now have three design principles for future experiments: favor mornings over evenings, favor subtraction over addition, and favor environmental design over discipline. These principles will make your next forty-three experiments dramatically more effective than your first forty-three.

Try this: Gather every experiment record you have created during this phase — whether in a journal, spreadsheet, notes app, or scattered across documents. If you have fewer than five entries, include informal experiments you remember running even if you did not record them formally. Set aside sixty to ninety minutes of uninterrupted time. First, read every entry without analyzing — simply re-familiarize yourself with the full body of evidence. Second, create a simple comparison table with columns for experiment name, domain, outcome (succeeded, partially succeeded, failed, ambiguous), and one column labeled "surprising observation." Fill in every row. Third, look across the table for patterns: Are certain domains more successful than others? Do successful experiments share structural features? Do failed experiments share common causes? Write down every pattern you notice, even tentative ones. Fourth, select the single most important pattern and formulate it as a design principle for future experiments — a concrete rule like "I should always pair a new behavior with an existing trigger" or "My experiments work better when I tell someone about them." Fifth, open your experiment backlog and re-prioritize at least three queued experiments based on what you learned. The review is complete when your backlog reflects the intelligence your review produced.