snapkin
The targets · Decision 6 of 14
Why your steps and workouts do not change your calorie target
Snapkin reads your steps and your active energy and shows them on the home screen, without a ring around either. By default it does not add them to what you may eat. A daily activity estimate is already inside your starting target. Adding a watch reading on top of that counts some activity twice. Since September 2026 you can choose the other way round: the estimate drops to the lowest level and the measurement is used instead.
Last updated .
- What the app does
- Shows your activity. Adds it to your budget only if you ask. By default a 15 km Sunday does not raise your target on its own. Under Settings › Active calories you can swap the built-in estimate of your usual week for the measurement of that day. If two weeks of active calories sit a whole level away from the level you chose, the app suggests another level. Nothing changes until you confirm.
- How sure we are Fair , Real, but small, indirect, or thinly measured.
- One living review, and its energy section is 8 studies and 270 people. They all point the same way. Those results cannot tell you how far off one workout on a current watch is.
The evidence for it
All three numbers are from one living review, a review that is kept up to date. Heart rate and energy estimates were tested in different studies and conditions. An accurate heart rate does not guarantee an accurate calorie count.
A starting suggestion during setup
If you allow health access, setup reads the previous 14 completed days in your time zone. Today is left out. We average the days with step readings. We need at least 7 recorded days before we suggest an activity level. Missing days are left out, not counted as zero. We show both the average and the number of recorded days. The seven-day minimum is our own product rule, not a tested medical limit.
| Average steps per recorded day | Suggested level |
|---|---|
| Under 5,000 | Desk job |
| 5,000 to below 8,000 | Lightly active |
| 8,000 through 12,000 | Active |
| Over 12,000 | Very active |
Active energy readings are averaged separately, and shown with their own count of recorded days. They do not pick the activity level, and they do not add calories to the budget. They can make the app ask you to check your level, as explained below. Only the activity category you confirm is saved to your account. The readings stay on your device. The confirmed category goes into the daily calorie estimate during setup. This is a starting estimate. It is not the watch adjusting your budget day by day.
Connecting a health store does not guarantee there is data. Health Connect's documentation describes gaps in its totals, how it removes duplicates from several sources, and its default 30-day history window. HealthKit does not say whether read access was denied. If the history is empty or cannot be read, you answer the activity question yourself.
Why we show both histories, and do not invent a calorie cut-off
Below the activity choices, you can open a panel with daily charts for steps and active energy, lined up day by day. Each has its own scale, average and count of recorded days. Tapping a date shows both readings. A missing reading is not the same as a recorded zero. The suggestion is based on steps only, and you can change it. If you cycle, swim or lift weights, keep that training in mind when you confirm it.
Mitchell et al. (2024) reviewed 166 studies of how many calories weight training burns, including 31 that used wearable monitors. The estimates varied widely. Even laboratory methods struggle to measure the short, hard bursts of effort in lifting. That is a reason to show the extra reading. It is not a reason to set a calorie cut-off for our activity categories. Steps miss activity too. Neither chart can tell what a person's energy needs are.
So we do not add steps and active calories together. We never move you to another category by ourselves. That leaves a gap for exercise that is not walking. The charts, the question you can edit, and the check below are how we handle that gap. We do not claim to have a tested way of combining the two. Only the confirmed category is saved.
After setup, steps only suggest going up
Once a day the app reads the same 14 completed days again, on your phone. If your average steps now fall in a higher band than the level you saved, and that level would change your target by at least 50 kcal or 5 g of protein, the dashboard asks you to review your goals. Nothing changes until you confirm it.
Steps never suggest a lower level. Steps miss cycling, swimming and strength training, so fewer steps can just mean a different kind of exercise. Following that would lower the target for people who train. We also stay quiet if you typed your own target, or if you chose to count measured movement instead. If you tap "Not now", we do not ask about that level again for 90 days. Your steps still never leave your phone. Only the level they point to is used to work out the new target.
Active calories can suggest either way
The same daily check also looks at your active calories. Each activity level is a multiple of your resting burn, so the app works out how many active calories that level should mean for your body. If your average over the last 14 completed days is a whole level higher or lower than the level you saved, the dashboard asks you to review your goals. The same 50 kcal rule, the same “Not now” and the same 90 days apply. We need at least 7 days with readings. The readings stay on your phone.
Here a lower reading can count. A watch sees the bike ride and the swim that steps miss, so fewer active calories is not just a different kind of exercise. But active calories never suggest a level below the one your steps point to. Otherwise saying yes to one suggestion could bring back the other.
The sum is energy balance, not a guess. A day’s burn has three parts: your resting burn, digesting food, and moving. Digesting food is about a tenth of the day. That is the split Murakami et al. (2019) use to get movement out of a doubly labelled water total. Apple’s white paper calls active calories the calories burned above resting. Android’s Health Connect describes them as active energy “excluding basal metabolic rate”. So active calories are the moving part. For a level with multiplier m, that is (0.9 × m − 1) × resting burn.
| Level | Multiplier | Man | Woman |
|---|---|---|---|
| Desk job | 1.35 | 380 | 270 |
| Lightly active | 1.5 | 610 | 450 |
| Active | 1.65 | 850 | 620 |
| Very active | 1.8 | 1,090 | 790 |
A desk job is not zero. It still means walking to the car, to the kitchen and around the shop. At our lowest level, that is about a fifth of your resting burn.
Why a whole level, and not the nearest one. In the four-watch test further down, the best watches were off by about a third on average. For the man above, a third of 600 active kcal is about 200. One level is 240. So a reading halfway between two levels tells us nothing. A reading a whole level away is still not proof. That is why the app asks, and nothing changes until you confirm.
What this does not know. No study has tested what a current watch reads over ordinary days, against the best method. Older wrist trackers read too low over normal days. In the two recent lab tests further down, the Apple Watch read too high during exercise. So the check can be wrong in either direction. The level you pick is the one we use.
The other arrangement, if you choose it
Since September 2026 there is a setting for this. It is off unless you switch it on. It is one choice with two halves, because either half on its own is wrong. If you choose measured movement, your activity level drops to the lowest one the calculator offers, and your watch’s reading for that day is used instead. If you choose the estimate, the level goes back. There is no way to count both.
It does not remove the double counting. It makes it smaller. The lowest level is 1.35 × BMR, not 1.0. A person who is not stuck in bed still moves, and a watch does not see all of that movement. Take an 80 kg, 180 cm, 35-year-old man as an example. The level alone is the difference between a 2,350 and an 1,800 kcal target. And roughly 614 kcal of assumed daily movement is still inside the lower one. We have not tested whether adding a whole watch reading on top of that avoids double counting. We are not claiming it does. It is a choice we let you make, with the sums shown. It is not a result we can cite.
The measured figure is one number per day: what your watch reported the last time you opened the app. If you opened it once in the morning, that figure covers part of the day, and it stays that way. Your steps are never sent to us at all.
What the review found
Lambe et al. (2026), is a systematic review and meta-analysis in npj Digital Medicine that is kept up to date. It pooled 82 studies covering 430,052 people across 14 things an Apple Watch measures. On calories burned it is blunt. Margins of error “were often large, both during exercise and at rest”. All six studies that worked out a MAPE reported values of 20% or higher in at least one test. There was no clear pattern by how hard the exercise was. So there is no single correction that would fix it.
Is one brand of watch good enough?
We checked whether an Apple Watch is good enough to use, even if other brands are not. The research does not show that. Every brand that was tested got calories wrong by a lot. The brand that did best changed from one study to the next.
Kostrna et al. (2026) tested four watches at the same time. They had 44 to 52 adults per watch, during an indoor cycling test. The watches were an Apple Watch Series 8, a Fitbit Sense 2, a Samsung Galaxy Watch 5 and a Garmin Forerunner 955. Each was compared with a mask that measures the air people breathe. This is a preprint. Other scientists have not checked it yet.
Le et al. (2022) tested three watches on 20 young adults walking and running outside. There the Huawei Watch GT 2e did best, with an average error of 10 to 12%. The Apple Watch Series 6 was at 20 to 24%, and read too high. The Garmin Fenix 6 was at 22 to 32%. Huawei paid for that study.
Reviews of all brands say the same. Fuller et al. (2020) looked at 158 studies of nine brands. Their summary: “For energy expenditure, no brand was accurate.” Steps did much better than calories in the same review.
An activity level is about whole days, and there is very little research on whole days. In the same review, only 3 of 17 calorie comparisons in daily life were within 10%. Murakami et al. (2019) gave 19 adults 12 devices for 15 normal days. They measured the real calories burned with doubly labelled water, the most exact method there is. Almost every device read too low. None of the wrist trackers showed who burned more and who burned less. These were older trackers, and there was no Apple Watch among them. We found no study that tested an Apple Watch over whole days in this way.
An hour’s walk: active calories or total calories?
There is no single calorie cost for an hour’s walk. Body weight, speed, slope and terrain matter. Apple’s white paper on how the watch counts calories separates active calories, which are on top of what you burn at rest, from total calories, which include what you burn at rest. A reading of around 200 active kcal can be reasonable. Its size alone does not show whether the watch is accurate.
The 2024 Adult Compendium lists level walking at 2.5 mph as 3.0 MET, 2.8–3.4 mph as 3.8 MET, and 3.5–3.9 mph as 4.8 MET. A MET is a unit of effort: 1 MET is what you burn sitting still. The examples below use an 80 kg adult as an example, not a measurement of a real person.
| Walking speed | Active kcal | Total kcal |
|---|---|---|
| 4.0 km/h · 3.0 MET | 160 | 240 |
| 4.5–5.5 km/h · 3.8 MET | 220 | 300 |
| 5.6–6.3 km/h · 4.8 MET | 300 | 380 |
The Compendium's rule is 1 MET ≈ 1 kcal/kg/hour. So total kcal ≈ MET × kg × hours, and active kcal ≈ (MET − 1) × kg × hours. The resting part here is 80 kcal for the hour. This is a rough rule for people in general, not Apple’s personal calculation.
So in this example, around 400 total kcal fits a brisk walk, and around 200 active kcal fits a moderate one. Neither is the value for every hour’s walk.
The evidence against, and the limits of what we are relying on
This is one review, and its section on calories is small. Only 8 of the 82 studies tested calories burned. They had 270 people in total, 63% male. The review notes that the people were “predominantly young physically active adults”. Five of those eight tested an Apple Watch Series 2 or older. Newer watches and newer software may well be better. This review cannot tell you that they are not.
The studies that compare brands are small. Each tested one kind of exercise, mostly on young, healthy adults. None tested a current watch over whole ordinary days. So a newer watch may be good enough, and we cannot rule that out. If a study shows it, we will look at this again.
And our default has a real cost. Somebody who really does run 15 km on a Sunday gets a target that ignores it. We still think a plain number you adjust on purpose is better than a nicer-looking one worked out from an estimate that can be wrong by half. That is why the default stays. The setting above is the answer to that complaint. It is off unless you ask for it. The screen says that measurements are estimates too. And it makes the double counting smaller rather than claiming to remove it.
There is a second argument we find convincing and are not citing, because we have not read the original studies. It says the body partly makes up for exercise by burning less elsewhere. If so, even a perfectly measured workout does not raise your total daily burn by the same amount. If that holds, it strengthens the case above. We will add it here properly, with numbers, when we have read the studies rather than summaries of them.