You weighed 78.4 kg on Friday and 79.6 kg on Monday, and you have not gained 1.2 kg of anything in three days. What you have done is eat more salt, more carbohydrate and more food than usual, and your body is holding water accordingly, the way it has for the entire history of your species.

You know this. Everyone knows this, in the abstract, at a comfortable distance from a scale. And almost nobody acts on it, because the number is concrete and the explanation is not, and at seven in the morning the concrete thing wins. It always wins. That's why it's concrete.

What the daily number is actually made of

Day-to-day fluctuation is dominated by things with no relationship whatsoever to fat or muscle:

Water bound to carbohydrate. Glycogen stores with roughly three grams of water per gram. A high-carb weekend moves the scale a kilo entirely on its own, and gives every gram back by Wednesday.
Sodium. A plain fluid shift. Resolves in a day or two, having ruined a morning.
Digestive contents. Food in transit has mass. A high-fibre day shows up on the scale, briefly and misleadingly.
Training. Hard sessions cause inflammation and fluid retention in exactly the muscles you worked. The scale punishes you for the workout. Sit with the injustice of that.
The menstrual cycle, where applicable: a predictable monthly swing large enough to swallow a week of careful dieting whole.
Sleep and stress, through cortisol and its effects on fluid balance, because of course the bad night gets to count twice.

Stacked together, typical day-to-day variation is one to two kilos. A reasonable rate of actual fat loss is around 0.5 kg per week.

Do the arithmetic once and let it hurt: the noise is two to four times the size of the weekly signal. Judging progress from one morning's weight is judging the climate from this afternoon. People do both, constantly, with conviction.

What a moving average actually does, and what it can't

The standard fix is a moving average, seven days or fourteen: look at the average of the last week instead of today's number, and the daily fluid noise mostly cancels, being roughly symmetric around the truth.

This genuinely works, and it's why every serious tool draws a trend line rather than a dot. It is not, however, absolution, for two reasons:

A seven-day average still carries real error. Averaging seven noisy readings cuts the noise by a factor of about two and a half. Not to zero. If your daily noise is ±1.5 kg, the weekly average still wobbles by a few hundred grams, which is the same size as an entire week of actual progress. The average is better. It is not done.

Some noise isn't daily. A menstrual cycle, a holiday, a bad week of sleep, starting creatine. These move the average itself, not just the days inside it. A seven-day window cannot average out a fourteen-day pattern, no matter how sincerely you stare at it.

So how long do you actually have to wait?

For body weight: three to four weeks is the shortest window in which a trend deserves your trust, and about twelve weeks is where you can finally see the shape of a training block instead of the shape of your last holiday.

That is an uncomfortably long time to wait for feedback, and it is precisely why the people who weigh daily and interpret daily end up rewriting their diet every ten days, and thereby never find out whether any version of it worked. They ran nine experiments and finished none. I have watched people do this for years. I have done it.

The workable compromise: weigh daily, because more data points make the average better. Look only at the trend. Act only on four weeks of it. Three verbs, three different frequencies.

Weight is one line, and one line is ambiguous

The other reason to stretch the window: on its own, body weight cannot tell you what happened. Flat weight with rising strength and a shrinking waist is an excellent outcome that a scale-only view files under failure.

Put strength, training consistency, measurements and intake on the same twelve-week timeline and the ambiguity mostly dissolves:

Weight flat, strength up, measurements down → recomposition. Touch nothing.
Weight down, strength down, intake far below maintenance → cutting too hard.
Weight up, strength flat, intake high → gaining more than you need to.
Everything flat for twelve weeks → now you've earned the right to change something.

Not one of those verdicts is available from any single line. The information was always in the correlation. It usually is.

Doing it in OneRep

The progress view puts strength and volume trends, weekly consistency, measurements and check-ins on one twelve-week timeline, next to what you actually ate. Add your own metrics if there's something else you want on the same axis.

The twelve-week default is not a design flourish. It is a deliberate obstacle between you and overreacting. A rough Monday is just a rough Monday, and by Thursday you won't remember it. Which, on this one subject, is the correct amount of memory.

Try it: Free, no trial window.

Compare: OneRep vs MacroFactor · all comparisons