There's a post I keep seeing, written in almost the same words every time. Someone is seventy pounds down. The calorie target that carried them through the first sixty worked like a machine: eat the number, watch the scale drop, repeat. Then, somewhere past the halfway mark, the scale just stops. Same food, same logging discipline. A month of nothing. And the question underneath the frustration is always some version of: why did my number stop working?
Close behind it, two cousins. A smaller woman, around 5'3", asking whether these calculators are even accurate for petite people, because the number they hand her seems impossibly low next to what everyone else gets to eat. And a well-meaning commenter telling a beginner to track his TDEE every day, as if it were a stat on a watch face.
All three are the same misunderstanding wearing different outfits. The number didn't stop working. It was never the kind of number most people think it is, and once you see what it actually is, the stall, the petite problem, and the daily-tracking advice all untangle at once.
The math scales down fine. The advice doesn't.
Start with the accuracy question, because the answer is not what you'd expect. Nearly every mainstream calculator runs the same equation under the hood, Mifflin-St Jeor: ten calories per kilogram of body weight, plus 6.25 per centimeter of height, minus five per year of age, then plus 5 for men or minus 161 for women. That gives your basal metabolic rate, which gets multiplied by an activity factor between 1.2 and 1.9. Nothing in that arithmetic breaks at small sizes. It scales down as smoothly as it scales up.
Run it for a 35-year-old woman at 5'3" and 130 pounds with a desk job and you get a BMR around 1,254 and a sedentary maintenance around 1,505 calories. Lightly active, about 1,724. Moderately active, about 1,943. Those numbers aren't wrong. They're just small, and small numbers get punished by advice calibrated on bigger bodies.
Look at what actually circulates as advice. A 500-calorie deficit. A hard floor of 1,500 calories. A friend's target of 1,800 that worked great for them. For a six-foot, 200-pound man burning around 2,950 a day, a 500-calorie cut is a comfortable 17 percent trim. For the woman above, it's a third of her entire sedentary budget. The 1,500-calorie "floor" people repeat as a safety rule sits at her full maintenance, so eating down to the floor cuts exactly nothing. And the borrowed 1,800 target could be a real deficit, dead-on maintenance, or a slow surplus depending entirely on whose body it lands in.
So when petite women report that the calculators feel inaccurate, what they're usually colliding with isn't equation error. It's that every number they encounter in the wild was shaped by and for people with several hundred more calories of headroom. The calculator's low output is the honest part. The surrounding advice is what lies.
Your target has your weight baked into it
Now the stall. Look back at the equation: your current weight sits right inside it, at ten calories per kilogram. That works out to about 4.5 calories of BMR per pound, and after the activity multiplier, every pound you lose quietly removes roughly five to nine calories a day from your true maintenance, depending on how active you are.
Per pound, that's nothing. Over a big loss, it's the whole story.
Take someone who started at 330 pounds and now weighs 260. At 5'6" and 40 years old, lightly active, her maintenance at the start was around 3,000 calories. At 260, it's about 2,565. Nothing went wrong with her metabolism; 437 calories of daily burn simply left with the 70 pounds, because a smaller body costs less to run. If she set a target of 2,400 back at the start, she was eating 600 under maintenance then, and she's eating about 165 under maintenance now. The diet didn't break. It succeeded its way into a stall.
The same mechanism explains the 5'3" version of the mystery. At 200 pounds and sedentary, her maintenance is about 1,886, so eating 1,800 loses weight, slowly. By 165 pounds, maintenance has slid to roughly 1,695, and that same 1,800 now gains. An identical plate of food flips meaning as the body eating it shrinks. Anyone who computed a target once, months and many pounds ago, is aiming at a spot where they no longer stand.
The fix is almost insultingly simple: recompute with today's weight whenever the scale has moved meaningfully, say every ten pounds or so. The friction is that most calculators make you re-enter everything from scratch each visit, which is exactly why people don't bother. It's the one thing I built TDEEfy around: your inputs stay saved on your own device, so a recheck after a weigh-in means changing one field, and you can see all five activity levels at once instead of guessing which dropdown you picked last time.
An estimate isn't a measurement, so you can't track it
Which brings us to the advice about tracking your TDEE daily. It sounds rigorous. It's actually a category mistake.
A calculator TDEE is an estimate, a statistical prediction fitted across hundreds of study participants and fed a handful of numbers about you. Give it the same inputs and it returns the same output forever. The only input that changes day to day is your weight, and daily weight is mostly water: an overnight two-pound swing from salt, carbs, or a hard workout moves the formula's output by all of 11 to 17 calories, depending on your activity multiplier. Meanwhile your actual energy expenditure genuinely does swing by hundreds of calories between a training day and a couch day. Punching your morning weight into a calculator every day gets you a noisy readout of nothing: a number that wiggles far less than the real quantity it's supposed to represent.
What you can track daily is intake and body weight, and that pairing is the real measurement. Eat a fixed number for three or four weeks and watch the trend line, not the individual weigh-ins. If you're holding steady at 2,000 a day, your true maintenance is about 2,000, whatever any calculator says. If you're dropping half a pound a week at 2,000, you're running roughly a 250-calorie daily deficit, which puts your actual maintenance near 2,250. The calculator's job was to hand you a sane opening bid. The scale's job is to correct it.
The three answers are really one answer. The number is an estimate of a moving target: a population-level guess that your own results refine, that shrinks as you do, and that can't be read off a screen each morning. Handled that way, the number that "stopped working" works exactly as well as it ever did. It just needed you to stop treating a snapshot like a subscription.