Methods
Every number Kept uses to decide how hard your week should be, with the source it came from — or an admission that it hasn't got one. Plenty of apps say "science-based". This is the page that lets you check.
Step one
Field tests, not lab equipment — you need a floor, a chair and a step. The battery follows ACSM and ACE conventions, and the balance between them follows the 2025 study PMID 40672648, which found that isotonic tests (push-ups, chair stands) track reference measures better than isometric holds, which ceiling out. That's why the plank is included but counts least.
Upper-body muscular endurance
Maximum consecutive push-ups with no rest and no pause between reps, full range — chest down to a fist or rolled towel, full elbow extension at the top, hips and shoulders in a straight line. Stop when you can't complete a rep or lose form for two in a row. The scoring table (CSEP) is built on standard push-ups for the men's column and modified, knees-down push-ups for the women's. Use the version your column was normed on, or the score won't mean what it says.
Lower-body muscular endurance
Sit in the middle of a chair, feet flat, arms crossed over your chest. Stand up fully and sit back down as many times as you can in 30 seconds. Any firm chair with a seat around 43-48cm will do — that is an ordinary dining chair, not special equipment. Not sure? Sit down and look at your knees: they should be bent at roughly a right angle with both feet flat on the floor. Avoid a sofa or an armchair (too low and too soft), a stool (no back, and usually too high), and an adjustable office chair unless you can set the height and keep it there between tests. No suitable chair at all? Skip this one. Your other three still score, and the app will say your lower-body result is missing rather than quietly counting it as zero.
Cardiorespiratory fitness
Step up-down on a 30 cm step at 96 beats/min (24 steps per minute) for 3 minutes. Sit down immediately, then count your pulse for a full 60 seconds. Enter that number.
Core endurance
Forearm plank, straight line from shoulders to heels. Stop the clock when the hips sag or pike. Isometric holds ceiling out, so this counts least toward your score.
Step two
A raw number means nothing on its own — 25 push-ups is a different achievement at 22 than at 62. Each result is placed against a normative table for your age and sex. Here is exactly which table, and how much we trust it.
Canadian Society for Exercise Physiology, CSEP Physical Activity Training for Health (CSEP-PATH) Resource Manual, 2nd ed. (2019) — the table ACE reprints and the same lineage ACSM carries, going back to the Canadian Association of Sport Sciences in 1987. Its five categories map one-to-one onto our five bands, so these are published cut-points with no interpolation.
Read this before you test: CSEP built the men's column on standard push-ups and the women's on modified, knees-down push-ups. Use the version your column was normed on, or your score won't mean what it says.
Two sourced anchors with an interpolated middle. Under 40: healthy adults 18–35, mean 33.0 ± 5.4 reps. That study reported no split by sex, so neither do we — inventing one would add precision the data hasn't got. Over 60: Rikli & Jones (n = 7,183), test–retest ICC .84 men / .92 women, validated against weight-adjusted leg press (r = .78 / .71).
Ages 40–59 are interpolated between those two anchors, because nothing published covers the gap. They are the least trustworthy numbers in the app.
This test replaced a 60-second bodyweight squat, which had no published norms anywhere. We'd rather use a validated test that needs a dining chair than an unvalidated one that needs nothing.
YMCA submaximal step test, recovery heart rate counted for a full minute while seated. Our cut-points sit within about three beats of the published table — the best-calibrated numbers we have.
No ACSM norm exists for the plank, so the bands come from Strand et al. (2014), n = 471 — the 20th, 40th, 60th and 80th percentiles of hold time, split by sex. Using percentiles directly means "average" is literally the middle of the distribution rather than a number someone liked.
Two caveats we're keeping: that sample is 20-year-old students, 23% of them NCAA varsity, so it runs high. And only the under-30 band is measured — older bands are scaled down by the documented decline of 10–15% per decade after 35. The direction is published; the exact multipliers are ours. This is why the plank counts least of the four.
Step three
Prescribing intensity as a share of measured capacity is the foundation of exercise prescription — %1RM for strength, %HRR or %VO2R for cardio. So your rep target for each interval is a fraction of the maximum you actually recorded, not a number attached to a level band.
Because the denominator is your own test, everything gets harder on its own when you retest and your maximum has gone up. Nobody has to move a threshold.
Our judgementThe fractions — 50% and 40% — are our judgement. Percentage-of-max prescription is standard practice; the right percentage for bodyweight circuits specifically is not settled the way %1RM is for strength.
Score 2 or below in a domain and its exercises are regressed — incline push-ups instead of push-ups. Score 5 and they progress — diamond push-ups, jump squats, plank jacks. Per domain, not globally, so a strong core isn't held back by weak push-ups.
Progression needs a 4 — "very good" or better in that domain. It sat at 5 while our push-up and plank tables were unverified, because progression means plyometrics and handing those to someone who tested above their real capacity is the one failure here with a joint at the end of it. Both tables are sourced now, so the margin has come off. An easy session costs you a session; a movement you weren't ready for costs more.
Three to five sessions a week, scaled to your score. Training frequency is a genuine dose-response lever; sessions per day is not, and nothing in these norms supports "fitter means train twice today". Adaptation happens in the recovery, which is why the answer to getting fitter is harder sessions rather than more of them.
The part most apps skip
Four tests drawn from four different populations, vintages and study designs — a 1987 Canadian population survey, a 2014 study of American college students, the YMCA manual, and a young-adult chair-stand paper with 81 participants. Your overall level multiplies them together with weights that sum to one.
That is an index we designed, not a validated score, and combining incommensurable scales is exactly the kind of thing a paper wouldn't get away with. The protocols are evidence-based and the individual bands mostly are. The number sitting on top of them is ours.
We tell you because the alternative is a fitness level that sounds measured and isn't. It's a useful index — it moves in the right direction when you get fitter, and it's the same index every time so your progress is real. It just isn't a fact about your body.
Measurable adaptation in muscular endurance and cardiorespiratory fitness typically appears after about four weeks of consistent training. Testing more often mostly measures how you slept.
References
Last reviewed 5 August 2026. Where a source is paywalled we say so rather than implying we read it.