Baseline establishment is the pre-intervention tracking period during which a user collects enough data on their normal state to interpret any later change as signal rather than noise. It is the step most users skip and most regret skipping.
Why it matters
Without a baseline, every post-intervention number is floating. A user on a new ashwagandha stack who sees a 7.5/10 energy score does not know whether that is a lift from their usual 6/10 or a drop from their usual 8/10 — because they never measured their usual. This is the single biggest reason self-experimenters draw confident conclusions from insufficient data, and it maps directly to the stack patterns covered in supplement stack mistakes to avoid.
How long a baseline should run
The useful baseline length depends on the outcome's variability:
| Outcome | Minimum baseline |
|---|---|
| Daily subjective scores (energy, focus, mood) | 10–14 days |
| HRV and resting heart rate | 14 days for a stable rolling average |
| Sleep efficiency | 14 days, avoiding travel weeks |
| Training performance | 3–4 sessions at each target workout |
| Blood biomarkers | 1 draw at baseline; ideally a second 2–4 weeks later to confirm stability |
Shorter baselines are usable but produce noisier comparisons. A 3-day baseline can be deeply misleading if any of those 3 days had unusual sleep, training, or stress.
What a good baseline captures
Effective baseline tracking is boring. It captures:
- The daily value of each target metric.
- A stable moving average window so the reference point is smoothed.
- Any obvious confounders (travel, illness, alcohol spike, big training week) with a one-word tag.
- The subjective proxies or objective proxies that the user actually cares about — not every metric the device offers.
Once the baseline is stable, the active intervention begins. The n-of-1 experiment framing then compares post-baseline block means against the baseline block mean.
When baselines drift
Real life drifts. A user who set their HRV baseline in February and is evaluating a supplement in August has a stale reference if they gained 10 pounds, started a new job, or changed sleep schedule. Unfair re-baselines rolling references every 28 days by default, so the comparison is always against recent-past self, not long-ago self.
How this appears in Unfair
On first launch, Unfair guides the user through a 14-day baseline window before recommending any active stack changes for goals that depend on trend-level data. Baselines are visible on every review chart as the reference line, and the ranked output holds conservative weighting during the baseline period to avoid reacting to incomplete data.
Clinical safety note
A "baseline" is a reference for self-comparison, not a clinical normal. Out-of-range lab values or symptom-level biomarker readings should be discussed with a clinician regardless of where they sit relative to the user's personal baseline.