Reserve, the daily score

Reserve is one number from 0 to 100 that summarizes what you have to work with today. It comes from the eight questions in the check-in, each a slider from 1 to 10.

1. Eight questions. Sleep quality last night. Energy. Emotional reserve. Physical state. Sense of calm. Presence with family and friends. Fulfillment in work. Recovery time today.

2. Each answer becomes a 0 to 100 contribution: value × 10. A slider you leave untouched counts as 5, so 50.

3. The eight questions roll into six domains. A domain's score is the average of its questions.

4. Reserve is the equal-weighted average of the six domain scores, rounded to a whole number.

Why every domain counts the same

No data yet justifies weighting one domain above another, and a secret weighting formula is exactly the part of consumer wearable scores that does not hold up. Equal weights are transparent and checkable. The weights change only when real data earns a change, and that is a decision made in the open, not a tuning knob.

What Reserve is, and is not

Reserve is a daily composite readiness index, modeled on how elite athletic daily-wellness monitoring works: short self-report items, aggregated into one readiness number, computed entirely on your phone. It is labeled in the app as a directional indicator, because that is what it is today. It is not yet a validated clinical scale. The validation plan is at the bottom of this page.

Reserve means the most within one person over time. Your Tuesday against your Thursday is signal. Your 62 against a colleague's 68 is not the point, and it is not how the product uses the number. Every serious claim Endure makes is a within-person claim.

Levers, what moves your Reserve

A Lever answers one question from your own data: which conditions in my work measurably raise or lower my Reserve? Each condition you tag in a check-in (the chips) becomes a candidate Lever.

1. Split your logged days into the days you tagged the condition and the days you did not.

2. Average your Reserve on each side. The difference is the Lever: mean(with) − mean(without).

3. Wait for enough days. A Lever needs at least 3 days with the condition and 3 without it. Below that it reads "building data" and is never shown as a result. The earliest a first Lever can appear is your sixth check-in.

The rules that keep a Lever honest

Why within-person is the whole trick

Each clinician is their own control. Comparing your days with a condition to your days without it cancels out everything that differs between people: personality, baseline mood, specialty, how you use a scale. That is the cleanest cause-and-effect signal available without running an experiment on someone, and it is a recognized approach, called single-subject or n-of-1 method. The minimums and the constant rule are grounded in published standards, listed in the bibliography.

The aggregate, when an organization runs Endure

When clinicians in an organization choose to share, the Conditions Report does not pool everyone's raw days, because one heavy logger would then dominate the numbers. Instead:

1. Each person's own within-person difference is computed first, on their phone's terms, exactly as above.
2. Those personal differences are averaged across people, one vote per clinician.
3. Nothing is shown for fewer than five clinicians. A small group rolls up into its parent group or is left out.

So an organization reads "on average, across clinicians with enough data, inbox days cost about 12 points of Reserve." Never an individual, never a group under five.

What the numbers can and cannot say

• They are descriptive. A Lever says exactly what it computes: your average on days with a condition against your average on days without it.
• They are about you, over time. Not a ranking against anyone else, and not a diagnosis of anything.
• They need days. More logged days tighten every estimate. A single bad week is signal, not failure, and nothing in the app counts consecutive days or resets.
• Sample data never touches them. The sample month you can switch on to explore the screens is fictional, lives only in memory, and disappears the moment you log your first real day. It is never sent anywhere.

Validation, and where a statistician comes in

For the daily product, the analysis is transparent within-person arithmetic, and it does not need a statistician to be valid, because it claims only what it computes. Proving that Reserve tracks something real is a different job, and that job is under way.

• A validated monthly anchor. A twelve-item monthly instrument for clinician working conditions is going through its own validation study. Once it is validated, it joins the app as a monthly check-in alongside the daily score.
• Convergence. Each person's average daily Reserve in a month is then compared with their monthly score. If Reserve tracks a validated measure of the same thing, that is convergent validity, and that analysis is run with a biostatistician.
• Any published claim. Multilevel models (days nested within clinicians), reliability, missing days, and measurement reactivity are handled with a statistician before anything is published.
• Re-weighting. If the six domains ever stop counting equally, that is a data-driven decision a statistician leads, not a product hunch.