Overview
Condition Score is a daily 0–100 summary of the body context Grio can observe for one local date. It answers a narrow question: compared with how your own body has been reading lately, what does today’s evidence look like?
It is deliberately not a diagnosis, not a judgment of training readiness, and not a population norm. It also does not independently change your nutrition targets. It exists so the coach and the app share one grounded number instead of each guessing from raw sensor readings.
The number is worth a glance every morning for one reason: it is graded against you. A resting heart rate of 58 bpm is unremarkable in the abstract and can be a clear signal if you normally read 51.
How it’s calculated
What goes in
Grio combines only the evidence that is actually present for the date:
| Contributor | Where it comes from |
|---|---|
| Heart-rate variability and resting heart rate | Privacy-reduced Apple Health projections, compared with your own baseline |
| Sleep | The night Apple Health recorded, or the sleep you reported in a Check-In when it did not |
| Condition Check-In | Your sleep-quality rating and self-reported body state |
| Previous-day activity load | Recorded workouts and movement, relative to your recent load |
| Respiratory rate and wrist temperature | Supporting signals only |
The cardiovascular recovery axis carries the most weight, followed by sleep, then the Check-In, then previous-day activity load. Heart-rate variability and resting heart rate share the recovery axis: when a device reports only one of them, a watch that writes resting heart rate and no variability, that reading is the axis and carries the whole share, because handing the difference to sleep and self-report would let the only physiological evidence of the day be outvoted.
Respiratory rate and wrist temperature are anomaly signals, not independent recovery evidence. They can only deduct, only when elevated beyond your personal band, and only alongside core recovery signals. They never push a score up.
The algorithm in plain terms
Each contributor is first translated onto a personal-deviation ruler: not “is this good?” but “how far is this from your own usual, measured in your own day-to-day spread?” That is what makes the score comparable across people who have nothing physiologically in common.
Every contributor is then oriented so that positive always means better condition, and passed through one continuous saturating curve before the contributors are combined. Two properties of that curve matter to you:
- Grading is monotone: unusually favourable evidence can never score below ordinary evidence. An unusually long night is not penalised merely for being unusual.
- The curve saturates: one extreme reading gets a large but bounded effect instead of unlimited influence over the day.
Sleep is the exception to pure personal grading. Sleep need is individual, but a short night is short however little you usually sleep, so sleep is anchored to an absolute curve: roughly seven and a half to nine hours sits on the plateau, and both shorter and much longer nights fall away from it, with your personal baseline adjusting that grade rather than defining it.
Two safeguards bound the low end. A severely low score requires corroboration from independent contributors; a single struggling signal can make only a bounded deduction rather than dragging the whole day to the floor. And one struggling physiological system is graded once, not deducted again through a separate weakest-link rule.
The rolling baseline itself is screened, so a single implausible artifact, such as a sensor glitch that reads as an enormous HRV spike, cannot inflate your normal spread and hide real movement for the rest of the window. That screening may remove rare artifacts but never a sustained change, and it never looks at the target day as part of its own baseline.
Finally, a separate comparison can colour the result. The Sustained Recovery Trend asks whether HRV or resting heart rate has stayed meaningfully different over the last seven local dates versus the preceding, non-overlapping 60. When that longer comparison reports a concern, Condition takes a small bounded bounded deduction, because a 30-day adaptive baseline will eventually absorb a long decline and stop noticing it.
Edge cases and fallbacks
- You slept without the watch: the Check-In’s reported sleep takes over the sleep contributor. Missing values stay missing; nothing is filled in as zero.
- No mature health baseline yet: a completed Condition Check-In can produce the displayed numeric score from self-report evidence alone.
- Only one recovery metric: the available reading carries the whole cardiovascular axis rather than being renormalised away.
- A single overnight variability sample: one sample is thin evidence for a whole night, so it is worth less than a night the watch sampled repeatedly.
- Nothing readable at all: Grio shows no score rather than inventing one.
How to read your score
The app labels three levels: low, steady, and strong. Their cut points sit near the lower and upper quartile of the score distribution the grading curve actually produces, so all three are reachable on ordinary days. Reading a 71 as “71% healthy” is the wrong frame; read it as “steady, and a little below where I have been sitting.”
Two common misreadings are worth naming:
- Higher is not always the goal: the score summarises observable body context. A very high score after four rest days is not a training achievement, and a lower score during a hard training block is often exactly what a hard training block looks like.
- It is not a verdict on your day: Your own current report outranks the wearable evidence. If you feel fine, you feel fine. Grio will not contradict you, and it will not infer a symptom you did not report.
What to do with it
A useful score changes one decision, not your whole day:
- Steady or strong, and you had planned a hard session: proceed, and let the Target Strain Zone rather than this number set the size of it.
- Low, with sleep as the driver: the practical lever is tonight, not this morning. Protect the bedtime rather than adding a recovery ritual.
- Low, with the recovery axis as the driver, and you feel fine: treat it as information, not instruction. Grio will not silently rewrite your Daily Plan or your goal on the strength of a wearable deviation.
When the coach brings it up, it can name which contributor moved the score and can offer one small recovery-oriented action. What it cannot do is change a committed plan or a durable target on its own. That stays an explicit decision you make.
Troubleshooting & FAQs
- Why did my score drop when I feel completely fine?
- The score reads measured evidence, not how you feel, and your own report outranks it. Complete a Condition Check-In and Grio will fold your answer into the same score. A Check-In alone can produce the number. If the drop looks wrong, check whether your watch recorded the night at all: a missing night changes which contributors were available, and that alone moves the score.
- Can I compare my Condition Score with someone else's?
- No, and the comparison has no meaning. The score is graded against your own rolling 30-day baseline, so a 70 describes your day relative to your usual, not a shared standard. Two people with identical HRV can have opposite scores.
- Does a low Condition Score mean I should skip my workout?
- It is not a training-readiness verdict and it is not permission to train or rest. It is one piece of context. Grio's separate Target Strain Zone is the surface that speaks about the day's training load, and Condition is evidence for that target rather than the target itself.