Personal science platform
Run rigorous self-experiments — randomised designs, Bayesian analysis, causal inference — and discover what actually works for your body and mind.
Free · No credit card required · Your data stays yours
Caffeine timing · 28 trials
Probability morning caffeine improves focus. Average effect +1.8 points (HDI: [+0.9, +2.6]).
Process
Choose a template or build from scratch. Set your hypothesis, pick conditions, and define what you're measuring. Takes five minutes.
We generate a randomised condition schedule. Log how each day went — under a minute per entry.
After enough trials, analysis updates automatically. Plain-English summary plus full Bayesian output if you want the detail.
Capabilities
A/B, crossover, and multi-arm protocols with a randomised schedule. No statistics degree required.
Posterior distributions, credible intervals, and probability-of-improvement. Honest uncertainty, not just a p-value.
Draw your causal graph, adjust for confounders, and get doubly-robust AIPW estimates that hold up even when the model is wrong.
Connect Oura, Garmin, or Strava. Outcomes fill in automatically from your ring or watch.
Opt into pooling and your anonymised results contribute to broader effect estimates across everyone running the same template.
One-click public link for any finished experiment. No account needed to read.
Spreadsheets don't randomise your conditions. They don't account for carry-over effects. They give you a mean and your own wishful interpretation.
Steady Practice is built on the same principles as clinical trials: pre-registered hypotheses, randomised designs, and effect estimates with honest uncertainty. The only difference is the sample size is one — and that one is you.
Pricing
Start free. Upgrade when you need more.
Free
Pro
Most popularNo. Every result comes with a plain-English summary. The Bayesian numbers are available if you want them, hidden by default.
Most crossover designs run 3–6 weeks — enough trials to see a real signal. You set the target; the app tells you when you have enough data.
Yes. Your data lives on this server and is never sold or shared. Export everything as JSON or CSV from Settings at any time.
When you opt in, your anonymised results feed community posteriors — broader effect estimates from everyone running the same template. Opt out per-experiment, anytime.
Your data stays yours. Existing experiments remain readable. You just can't create new personal ones beyond the 15-experiment free limit.
Free for 15 experiments. No credit card required.
Create free account