A subjective proxy is a self-reported number that stands in for a state no wearable or lab can measure well — perceived energy at 10 a.m., focus during a two-hour block, anxiety at bedtime, joint comfort after training. It is the primary evidence surface for nootropic, adaptogenic, and mood-adjacent stacks where a blood draw would miss the outcome the user actually cares about. A subjective proxy is not a feeling; it is a number captured on a fixed scale at a fixed time of day, logged often enough that a trend emerges.
What qualifies as a subjective proxy
A usable subjective proxy must be all three of:
- Defined. A 1–10 score is not a proxy until "what does 7 feel like" has a written anchor the user will reuse.
- Timed. Morning energy, afternoon focus, and pre-bed anxiety are different proxies and should not be averaged together.
- Logged quickly. A proxy that takes two minutes to capture will be skipped; one that takes five seconds will be filled every day. This is why the app invests in fast entry paths.
A score without a written anchor drifts. A score without a fixed time of day absorbs too much circadian noise. A score that takes effort absorbs selection bias — bad days get skipped and the trend looks rosier than it was.
Common subjective proxies by goal
| Goal | Subjective proxy | Scale | Logged when |
|---|---|---|---|
| Morning energy | Perceived energy on waking | 1–10 Likert | Within 15 min of first light |
| Focus | Deep-work quality during one fixed block | 1–5 | At the end of the block |
| Mood | Overall affect | 1–10 | Same time daily (commute, pre-dinner) |
| Sleep quality | Subjective restfulness, independent of hours | 1–5 | At wake-up |
| Anxiety | Somatic + cognitive anxiety composite | 1–10 | Pre-bed |
| Recovery | Muscle soreness + readiness to train | 1–5 | Pre-training |
| Joint comfort | Pain/stiffness in a named joint | 0–10 VAS | Fixed time or pre-activity |
| Digestive comfort | Bloating + discomfort composite | 1–5 | Two hours post-largest-meal |
The set should be small. Three well-defined proxies logged consistently outperform ten proxies logged sometimes.
Why subjective proxies matter despite their weaknesses
Most supplement outcomes that motivate purchases — "I want to feel sharper, calmer, more recovered" — are subjective by definition. No wearable currently measures focus, and the ones that claim to measure recovery are themselves correlated with subjective feel. Refusing to use subjective proxies means refusing to evaluate most of the reason people take supplements.
That said, they are noisy. Subjective proxies are vulnerable to placebo expectancy in the first weeks, to recall bias whenever logs are reconstructed from memory, and to mood spillover (a bad morning drags an unrelated focus score). The defense is structural — anchors, timing, and a baseline — not an attempt to be more honest.
Pairing with objective proxies
Subjective and objective proxies work best as a pair. When both agree on direction, the signal is considerably more trustworthy than either alone. When they disagree, the disagreement itself is informative.
| Subjective says | Objective says | Likely interpretation |
|---|---|---|
| Better | Better | Real effect; proceed |
| Better | Flat | Possibly expectancy; extend trial |
| Flat | Better | Response below perception threshold; may still be worth keeping |
| Worse | Flat | Nocebo or daily-life confounder; isolate before concluding |
| Worse | Worse | Real negative effect; stop and reassess |
This 2×2 is the default lens Unfair uses at the end of a review cycle before feeding a proxy trend into the recommendation ranking.
A numeric example
A user runs an ashwagandha trial for 8 weeks with pre-bed anxiety as their proxy. Baseline mean is 6.2/10 over 14 days (SD 1.1). Intervention mean is 4.8/10 over weeks 4–8. The 1.4-point drop is roughly a 1.3 SD shift — large enough to clear the noise even under conservative weighting, and it persists past the expectancy window, so it survives in the cycle review.
How this appears in Unfair
The daily check-in captures a fixed set of subjective proxies in under ten seconds with written anchors the user set once during onboarding. Review charts show subjective and objective proxies stacked on the same timeline so their agreement or disagreement is visible at a glance. Subjective-only conclusions are flagged and down-weighted in the ranking until a paired signal is available.
Clinical safety note
Subjective proxies can miss early warnings. A falling mood score that the user attributes to a supplement trial may actually be a treatable depressive episode; rising anxiety may be a medication interaction rather than an expectancy effect. Any persistent worsening on a subjective proxy should be treated as a reason to pause the stack and involve a clinician, not as a number to trend-line away.