The Half of the Equation Calorie Apps Can't Possibly Track

You've been eating at your maintenance minus 500 kcal for weeks. The math says you should be losing about half a kilo (around 1 lb) per week. The scale says otherwise. And you're starting to wonder whether the problem is you — your discipline, your logging accuracy, your metabolism — or whether the number you've been subtracting from was wrong to begin with.

It was probably wrong. But that's not the whole story. Even if you had the right starting number, it would have drifted.

Where your calories actually go

It's easy to think your daily calorie burn is something you can meaningfully steer with workouts, steps, and effort. But when you look at where the calories actually go, exercise turns out to be a much smaller lever than most people assume.

Your resting metabolic rate — the energy your body uses just to stay alive, running organs, maintaining temperature, replacing cells — accounts for roughly 60 to 70% of everything you burn. The thermic effect of food, meaning the energy cost of digesting and processing what you eat, adds around 10%. Non-exercise activity thermogenesis — all the movement that isn't deliberate exercise, from fidgeting to walking to the grocery store to adjusting your posture — accounts for another 15% or more, and substantially higher in active occupations. Formal exercise, the thing most people think of when they think of "burning calories," typically contributes roughly 5 to 15% for moderately active people, though it can run higher for those who train intensely and frequently.

The majority of what you burn every day is entirely invisible to you — and, critically, variable in ways you cannot directly control.

The invisible variable: non-exercise activity

Non-exercise activity thermogenesis — NEAT — is the wild card that most calorie calculations ignore entirely. In research by James Levine published in Arteriosclerosis, Thrombosis, and Vascular Biology, NEAT was found to vary between two people of similar size by as much as 2000 kcal per day at the extreme, driven primarily by differences in occupation and spontaneous movement. Two people who look identical on a TDEE calculator — same sex, same height, same weight, same age — can have actual daily expenditures that differ by nearly a full day's intake for a moderately active person.

How much of that gap does deliberate exercise explain? Almost none of it. In a Science study, Levine and colleagues overfed 16 non-obese volunteers by 1000 kcal per day for 8 weeks. About two-thirds of the increase in total expenditure came not from formal activity but from NEAT — fidgeting, ambulation, minor postural changes — and those individual differences in NEAT explained a roughly ten-fold range in how much fat each person actually gained. The same surplus, very different outcomes, because NEAT moved differently in each person.

NEAT also shifts when you restrict calories. When you eat less, your body unconsciously reduces movement. That can look like fidgeting less, sitting more, moving through your day a little more slowly, choosing the elevator rather than the stairs — none of it deliberate. Levine's work found that obese individuals sat on average about 2.5 hours more per day than lean counterparts, a difference worth roughly 350 kcal per day. This is where the calorie app starts lying to you without meaning to: when your intake drops, your expenditure drops with it, your true deficit shrinks, and the calculator has no way of knowing any of this happened.

Your metabolism is a moving target

Beyond NEAT, your resting metabolic rate itself adapts to sustained energy restriction. Hall and colleagues measured resting metabolic rate in contestants from the American reality show The Biggest Loser at baseline, at the end of the 30-week competition, and again six years later. At the six-year follow-up, contestants had regained more than 40 kg (about 88 lbs) on average, but their resting metabolic rate remained about 500 kcal per day lower than predicted for their body size and composition. Their bodies had adjusted their baseline burn downward in a way that outlasted the weight loss itself.

That's an extreme case, but the pattern shows up in ordinary dieting too. A British Journal of Nutrition review found the same in more typical weight-loss studies: energy expenditure often drops beyond what body-size change alone predicts, usually in the range of 3 to 15% of predicted resting energy expenditure, and it remains measurable even a year after the diet ends.

The practical implication: if you set up your calorie target months into a diet, that number was based on your metabolism at that point. Your metabolism has since adapted. Your "deficit" has shrunk — possibly to near zero — and the calculator hasn't updated. Your body has.

Why your TDEE calculator was probably wrong

When a calculator gives you a maintenance number, it usually starts with one of two equations: Harris-Benedict or Mifflin-St Jeor. Both use age, sex, height, and weight as inputs, then multiply by an activity factor the user selects. Mifflin-St Jeor is generally considered more accurate for contemporary populations. A validation study comparing it against measured resting metabolic rate in 337 adults found it unbiased at the group level — on average, not systematically high or low — but at the individual level, it was accurate to within 10% for only about 82% of people, with errors often running ±10% or more even before the activity multiplier is added.

Then come the activity factors. "Sedentary," "lightly active," "moderately active" — rough buckets that translate to multipliers like 1.2, 1.55, and 1.725, applied to a number that already carries meaningful individual error. There's no multiplier for "I walk fast and fidget constantly" versus "I sit at a desk and barely move between meals." Both might select "lightly active." Population averages are useful for building calculators. You are not a population average.

The result: your maintenance number was built on an estimate with ±10% individual-level error, multiplied by a rough category guess, applied to a body that has since adapted in ways no calculator can observe. "I'm eating 500 calories below calculated maintenance but the scale isn't moving" is not a mystery once you understand this. The starting number was wrong, and it's been getting more wrong every week.

Why your fitness tracker's calorie burn isn't reliable either

When you stop manually calculating TDEE, it's tempting to let a wearable fill the gap — letting an Apple Watch or Fitbit estimate what you're burning and eating back some or all of those calories. The problem is that wearables carry their own substantial error. In a JAMA Internal Medicine study comparing popular devices against metabolic chamber and doubly labeled water measurements, many wearables misestimated total energy expenditure by more than 200 kcal per day in everyday conditions, with error direction and magnitude varying by device and context.

The same pattern shows up in chamber testing: a study assessing 12 devices found that most differed significantly from a measured physical activity expenditure of around 529 kcal per day, with error direction depending on the device. And the wearable number sits on top of the same TDEE problem: your baseline estimate may already be wrong, and the activity figure adds another layer of uncertainty rather than solving it. The tracker gives you a number that feels precise and personal, but the underlying estimate can be large enough to erase a modest deficit entirely — or make a real one look bigger than it is.

The only instrument that captures both sides

The scale does something no calculator can: it integrates everything.

Every drop in NEAT, every degree of metabolic adaptation, every inaccuracy in your food label, every uncounted bite — all of it shows up in the weight trend over time. So does every error a fitness tracker makes estimating your exercise calories, and those errors are substantial: studies comparing wearable devices against gold-standard measurements found many devices misestimate total energy expenditure by more than 200 kcal per day in everyday conditions, with errors varying in size and direction by device and context.

The scale doesn't care what any of these instruments say. It measures the net result of the whole system. The input side of the equation is noisy too, as covered in Why Calorie Counting Is Less Accurate Than You Think → Food labels are allowed to be wrong. The output side is just as unreliable — arguably more so, because it shifts dynamically in response to your behavior in ways that are invisible to any tracking tool.

That's why reading the output rather than independently measuring both sides is, over a multi-week window, a more honest reflection of actual energy balance than any combination of food diary and TDEE estimate. The rate at which your rolling weight average changes — smoothed to reduce the day-to-day noise from salt, glycogen, and hormones — tells you what your net energy balance actually is, not what it was predicted to be. Calorintel reads that rate and converts it into an approximate daily calorie balance, giving you the one signal that captures both sides of the equation from a single daily number. For a step-by-step guide to reading and acting on that signal, see Calorintel in Practice.

Research referenced in this article

— NEAT variability between individuals of similar size, including sitting time differences and estimated daily expenditure gap, Levine et al., Arteriosclerosis, Thrombosis, and Vascular Biology, 2006

— Overfeeding study showing two-thirds of increased expenditure came from NEAT, explaining a ten-fold range in fat gain between individuals, Levine et al., Science, 1999

— TDEE component breakdown and definitions of NEAT, thermic effect of food, and exercise activity thermogenesis, Calcagno et al., International Journal of Environmental Research and Public Health, 2019

— Accuracy of Mifflin-St Jeor and other predictive equations for resting metabolic rate compared with indirect calorimetry in 337 adults, Frankenfield et al., Journal of the American Dietetic Association, 2005

— Persistent metabolic adaptation 6 years after The Biggest Loser competition, Fothergill et al., Obesity, 2016

— Systematic review of adaptive thermogenesis after weight loss in adults, Doucet et al., British Journal of Nutrition, 2022

— Accuracy of wearable devices for estimating total energy expenditure compared with metabolic chamber and doubly labeled water measurements, Dooley et al., JAMA Internal Medicine, 2016

— Accuracy of 12 wearable devices for estimating physical activity energy expenditure using a metabolic chamber, Murakami et al., JMIR Mhealth and Uhealth, 2019