Why is my TDEE calculator wrong

You put your numbers in, it said 2,400, and you have now eaten at 1,900 for six weeks with nothing to show for it. So either you are metabolically unusual or the calculator is wrong.

Almost always it's the calculator — and not because it was built badly. A TDEE calculator does something inherently approximate and then presents the result as a four-digit number, which is a formatting decision more than a scientific one. Here's every place the error enters, roughly in order of how much damage each one does.

The activity multiplier is doing most of the harm

A TDEE calculator runs in two stages. First it estimates your resting metabolic rate from height, weight, age and sex. Then it multiplies that by an activity factor you chose from a dropdown.

The second stage is where the money is lost, and it isn't close.

Those dropdown steps — sedentary, lightly active, moderately active, very active — are typically about 15% apart. On a resting rate of 1,600, one step is roughly 240 calories. Two steps is 500. And the labels are hopeless: "moderately active, exercise 3–5 days a week" says nothing about whether you spend the other nineteen hours on a building site or a sofa, which matters far more than the workouts do.

That last point has a number behind it. Energy spent on everything that isn't deliberate exercise — walking, standing, fidgeting, moving around a job — varies between people of similar size by a genuinely enormous margin, up to something like 2,000 calories a day at the extremes. No five-item dropdown can resolve that. You are being asked to hand-classify the largest variable in the equation, using a label, before you have any data.

The resting-rate formula is better than its reputation, and still an average

The equation underneath is usually Mifflin-St Jeor, and it deserves some defence. When the common formulas have been compared head to head, it predicted measured resting metabolic rate within 10% for more people than any alternative, with the narrowest error band. It is the right choice.

But "within 10% for most people" contains the problem. Height, weight, age and sex only ever accounted for part of the variation between individuals — a meaningful minority sit well outside 10%, and nothing on the input form tells you whether you're one of them. Ten percent of a 1,600 resting rate is 160 calories before the activity multiplier touches it.

The main reason is body composition. Two people of identical height, weight, age and sex can carry very different amounts of lean tissue, and lean tissue is what burns. The formula can't see it, because you didn't give it anything that would reveal it. Formulas that do use body composition are more accurate in principle and worse in practice, because they need a body-fat figure you probably don't have to better than a few points.

Your maintenance moved after you set the target

Even a perfect estimate goes stale, and it goes stale in the direction that hurts.

You're smaller. A lighter body costs less to run. Lose 10 kg and your maintenance drops by roughly 150–250 calories a day on size alone. The target that worked in month one is not wrong now; it was right then.

You're moving less. A sustained deficit quietly reduces unconscious daily movement — fewer steps, less fidgeting, lower session intensity — and none of it registers as a decision you made.

And there's a genuine metabolic adaptation, smaller than the internet says. Measured properly, resting expenditure after weight loss runs on the order of 100 calories a day below what body size alone predicts while you're still in a deficit, and closer to 50 once your weight has been stable for a few weeks. Real, worth knowing, and nowhere near a broken metabolism. Roughly half of what looks like permanent damage is just the effect of currently being in negative energy balance, and it resolves when that does.

The suspect nobody wants to name

Before you conclude the calculator is out by 400, check the other number in the comparison.

Under-reporting of food intake runs around 20–30% on average, and it is not lying — it's oil in the pan, bites while cooking, splashes of milk, and the weekend that didn't get logged. If your intake estimate is running 20% low, a calculator that was exactly right will look 400 calories wrong.

This matters for a specific reason: the two errors are indistinguishable from where you're standing. "My TDEE is lower than the calculator says" and "my logging is lower than my eating" produce the identical symptom — target hit, scale flat. The only way to separate them is a fortnight of genuinely complete logging, weekends included. It is dull and it settles the question.

The fix is to stop calculating and start measuring

Here's the reframe that makes all of this manageable: a TDEE calculator is not supposed to give you your maintenance. It's supposed to give you a starting point good enough to generate data.

The measurement is straightforward.

1. Eat at a consistent target for three weeks and log it the same way every day. The consistency matters more than the accuracy — a figure that's reliably 15% low still works, because the error sits on both sides of the comparison.
2. Weigh most mornings under the same conditions and take weekly averages. Individual mornings swing by a kilo or two on water alone, which is larger than a week's fat loss.
3. Compare week one's average to week three's. Every 0.5 kg of weekly change corresponds to roughly 500 calories a day of imbalance.
4. Your real maintenance is your logged intake, adjusted by that gap — expressed in the units of your own logging, which is exactly the number you need.

That result beats any formula, because it's measured on the only subject who matters and it's already calibrated to how you log. How many calories to maintain your weight walks the arithmetic through properly.

And then it goes stale again, because you keep changing — which is the real reason a static number never holds. A maintenance figure needs to be re-measured, not re-calculated, and that turns out to be the thing that separates trackers from each other. metrIQ vs MacroFactor is the honest version of that comparison: both start from the same idea that your intake data plus your weight trend beats any equation, and they part company over what happens next.

What Sophia does with this

She uses the formula exactly once — on day one, because she has nothing else yet — and then spends every week afterwards making it obsolete.

She holds your logged intake and your weight trend as separate pieces of evidence and reconciles them against each other. When they disagree, the equation loses: your actual data is the better source, and she moves your target to match what your body has been doing rather than what a population average predicted. That's also why she re-measures instead of asking you to re-run a calculator when you've lost 10 kg — the change in your maintenance shows up in the trend before you'd ever think to check.

The number you end up with isn't more precise than the calculator's. It's yours, which is a different and more useful property.