An n-of-1 experiment is a single-subject trial in which one person alternates between an intervention and a control (usually "on compound" and "off compound" or placebo) over repeated blocks, then compares their own outcomes across those blocks. It is the most honest way an individual can test whether a supplement actually does something for them.
Why it matters
Population-level RCT and meta-analysis results describe what happens on average, not what happens to a specific person. A compound with a small average effect size can be a strong responder drug for some users and inert for others. An n-of-1 design replaces "studies show" with "this is what happens in me," which is the only read that actually guides a personal stack.
How to structure one
A workable home protocol:
- Pick a single compound you can isolate. Stacking obscures attribution.
- Define one primary outcome in advance — an objective proxy like sleep efficiency, or a subjective proxy like a daily focus score.
- Block the trial — for example, four alternating 2-week periods: on, off, on, off. Randomize block order if you can keep yourself blind.
- Track daily with structured logs through a daily check-in at a fixed time.
- Hold everything else stable — same sleep schedule, same training load, same caffeine intake.
- Compare block means at the end rather than reacting to any single day, ideally across a moving average window.
This design is close enough to a randomized controlled trial structure to filter out most placebo and context noise, and it is what Unfair's review cycle is built around.
Common failure modes
- Running too short. Most chronic-outcome compounds need 4–8 weeks per block; 1 week on, 1 week off rarely resolves anything.
- Changing too much at once. A new compound, a new dose schedule, and a travel week do not combine into a legible trial.
- Ignoring the washout. Carryover effects from fat-soluble compounds (vitamin D, omega-3) extend for weeks after stopping, so "off" may not actually be off.
- Reacting to the first week. The honeymoon response is almost always a mix of placebo and novelty — see placebo expectancy.
How this appears in Unfair
Unfair's stack-cycle and review structure is explicitly designed to support n-of-1 workflows. The review screen compares block means across activation and washout periods and flags whether the difference is meaningful given the noise in the user's own data. This is the core loop behind the ranked output — the recommendation engine is nudged by confirmed personal responses, not just population evidence.
Clinical safety note
An n-of-1 experiment is not a substitute for clinical care, particularly for compounds with interaction risk or narrow therapeutic windows. Coordinate any self-experiment with a clinician when it overlaps with prescribed medication or a managed condition.