Why the scale lies most days

17 August 2026·8 min read

You did everything right for four days. You step on the scale and it is up 700 grams. Nothing you ate could have produced 700 grams of fat in four days, because that would have required roughly 5,400 surplus calories on top of everything you burned. So what actually happened?

Something perfectly ordinary happened, and it had almost nothing to do with fat.

What the number is actually made of

Your bodyweight on any given morning is fat, plus muscle and organs and bone, plus a large and constantly changing amount of water and undigested food. Only the first of those is what you are trying to change. Only the last of those moves quickly.

• Water follows carbohydrate. Your muscles and liver store carbohydrate as glycogen, and glycogen is stored wet: the figure usually quoted is around three grams of water bound per gram of glycogen. A well fed adult holds several hundred grams of glycogen. Run it down over a few low carbohydrate days and refill it with one big pasta dinner, and the scale can move a kilo overnight without a single gram of it being fat.
• Water follows salt. Sodium pulls water into circulation, and your kidneys take a day or two to settle the balance again. A takeaway meal, a restaurant dinner or a packet of chips can leave a visible mark on the next two mornings.
• Food in transit weighs something. A large dinner is still physically inside you when you weigh yourself, and so is the fibre slowing it down. A high fibre week and a low fibre week are not the same weight even at identical body composition.
• Poor sleep and hard training both raise water retention, through ordinary stress-hormone and repair processes. This produces one of the cruellest patterns in dieting: the week you train hardest is often the week the scale looks worst.
• Menstrual cycles move water in a repeating monthly pattern, commonly by one to two kilos. If you are comparing this Tuesday to a Tuesday three weeks ago, you may be comparing two different points of that cycle rather than two different amounts of fat.
• Even the scale itself contributes a little. Position on the mat, a different floor surface, a scale that rounds to the nearest 100 grams: none of it is large, but it is all added on top of everything above.

How small real fat loss is by comparison

A kilogram of body fat is roughly 7,700 calories. That single number is what makes the noise problem obvious once you put it next to a realistic rate of loss.

Ninety grams a day is the upper end of what most people should be doing. Your bathroom scale cannot see 90 grams through a kilo of water. Neither can you. Which means that on any individual morning, the change you are looking at is almost entirely made of things you do not care about.

The trap this sets

Here is the real damage, and it is not hurt feelings.

You are losing fat at a perfectly good rate. The water noise happens to point upward this week. You conclude that what you are doing is not working, so you change it: you cut harder, you add a session, or you decide the whole thing is hopeless and stop. Every one of those is a bad decision, and every one of them was caused by reading a signal that was never in the number.

It works in the other direction too, and that version is sneakier. A whoosh of water loss after a low carbohydrate week reads as spectacular progress, right up until the water comes back and you conclude you have “undone” it. Nothing was done and nothing was undone. The water left and the water returned.

The pattern to notice is that both mistakes come from the same root cause: treating a high-noise measurement as though it were a low-noise one.

What to watch instead

Weigh daily, but never react to a daily number. What you want is the trend: a smoothed line drawn through your weigh-ins, where one strange morning barely registers and a real change shows up as a direction you can believe.

The standard tool for this is an exponentially weighted moving average. Each day’s trend value is mostly yesterday’s trend value, nudged a little way towards today’s reading:

Suppose your trend sits at 84.0 kg and you step on the scale at 85.2 after a salty dinner. With a smoothing factor of 0.1, the trend moves to 84.12: it registers the reading without being hijacked by it. If tomorrow comes in at 84.9 and the day after at 84.7, the trend keeps easing up slightly. But if the readings come back down, the trend follows them back with equal calm. A single outlier moves it a little. Ten days pointing the same way move it a lot. That is exactly the behaviour you want from something you are going to make decisions with.

In practice:

• Weigh under the same conditions. First thing in the morning, after the toilet, before eating or drinking, and without clothes if you can. This does not remove the noise. It just stops you adding more on top.
• Weigh most days, not once a week. Weekly weighing sounds calmer, and it is worse. You get one sample, and one sample is entirely noise: you have no way to tell a bloated Sunday from a real gain. More readings are what make the smoothing work.
• Missing a day is fine. The trend does not need a perfect run. Four or five weigh-ins a week is plenty for the average to behave.
• Judge over a fortnight. Ten to fourteen days of trend is roughly the point where real change outruns the noise for most people. Below that you are reading tea leaves.
• Do not chase a single day. If the trend is falling, the plan is working, whatever this morning said.
• Take a second measure if the scale bothers you. Waist circumference at the navel, once a week, under the same conditions, moves for reasons that have nothing to do with your glycogen stores. So do photographs and how your clothes fit.

Why this matters beyond your mood

There is a technical reason to care about smoothing, not just an emotional one. If you want to measure your maintenance calories from your own data, the calculation divides your weight change by the number of days. A kilo of water noise on either endpoint becomes roughly 550 calories a day of error across a fortnight. That is more than most people’s entire deficit, and it will produce a target that is confidently, precisely wrong.

So the trend is not a comfort feature. It is the thing that makes the rest of the arithmetic worth doing at all.

How Cut handles it

Cut never headlines the raw number. The trend is what it shows, and the trend is what every calculation in the app runs on: your measured burn rate, your daily target, the projection towards your goal. A single bad morning cannot move your calorie target, because your calorie target was never built on single mornings.

It also refuses to deliver a verdict too early. In the first fortnight it says plainly that it is too early to read, rather than manufacturing an encouraging story out of water weight. That is deliberate. An app that congratulates you for noise has to be believed later, when it tells you something true.