Running Experiments

What Are Experiments?

Experiments let you test specific hypotheses about your energy patterns using structured baselines and data-driven effect tracking. Instead of guessing whether a change in your routine is working, you can measure it.

Creating an Experiment

1. Navigate to Experiments from the sidebar or bottom nav.
2. Click New Experiment.
3. Write your hypothesis — what you believe will change and why. Templates are provided for common patterns (testing whether a change reduces spending on a category, increasing gains, stabilizing a pattern).
4. Add one or more measures. Each measure defines what you're tracking (e.g., mean energy per entry for expenses in a specific category) and in which direction you expect change (increase, decrease, or stabilize).
5. Set the experiment duration (7–30 days).

Baselines

Before your experiment starts, the system computes a baseline from your historical data. This baseline captures your typical values (mean, standard deviation, sample size, trend) for the time window before the experiment.

If you don't have enough historical data, the baseline preview will warn you. At least 5 data points are needed for a meaningful comparison.

During the Experiment

Once active, keep logging your energy entries as usual. The system automatically computes daily snapshots comparing your current values against the baseline.

• The dashboard widget shows active experiments with progress bars and per-measure deltas.
• The experiment detail page shows a full chart of your values over time with the baseline marked as a reference line.
• A confidence indicator tells you how reliable the comparison is based on sample size and effect size.
• Data sparsity warnings appear if you haven't tracked enough data recently for a measure to be meaningful.

Stopping an Experiment Early

If circumstances change or you want to pivot, you can stop an experiment before it finishes. Stopped experiments still show their partial results and verdict. You cannot resume a stopped experiment, but you can re-run it with a fresh baseline.

Understanding Verdicts

When an experiment completes (or is stopped), each measure gets an individual verdict and the experiment gets an overall verdict:

• Hypothesis supported — all measures moved in the expected direction with a meaningful effect size (≥ 0.2 Cohen's d).
• Hypothesis partially supported — some measures met their target, but not all.
• Hypothesis not supported — no measures met their target direction with a meaningful effect.
• Results inconclusive — not enough data (fewer than 5 data points) to draw any conclusion.

The effect size (Cohen's d) is reported as negligible (<0.2), small (0.2–0.5), medium (0.5–0.8), or large (>0.8).

After the Experiment

• Re-run — creates a new experiment with the same hypothesis and measures but a fresh baseline from your current data.
• Sync to Budget — if the experiment tracked expense categories, you can push the results into your energy budget allocations.
• Archive — hides the experiment from the default list. You can show archived experiments with the “Show archived” toggle.

Tips for Good Experiments

• Change one variable at a time so you can attribute any effect to that specific change.
• Run experiments for at least 7 days to gather enough data.
• Keep tracking consistently — gaps in data reduce confidence in results.
• Use the baseline preview before starting to make sure you have enough historical data.
• You can run up to 2 active experiments simultaneously.