How Long a Weight Loss Stall Has to Last Before It Means Anything
The short answer
A scale that has not moved for less than two weeks carries almost no information, and even two weeks is thin. Before you can say a stall is real, you usually need about three to four weeks of weigh-ins compared as weekly averages — not single readings, and not the number on the scale this morning.
That sounds like a long time to wait. It is a long time. The reason is arithmetic, and once you have seen the two numbers side by side, the waiting is easier.
How much weight should actually come off in a week?
A kilogram of body fat holds about 7,700 kcal. A pound holds about 3,500.
So a deficit of around 500 kcal a day — a fairly determined one, sustained every day including Saturday — produces roughly 3,500 kcal a week, or about 0.45 kg. A gentler 300 kcal a day produces around 0.27 kg a week.
Spread across the week, that is 40 to 65 grams of fat a day. Your bathroom scale probably reads in increments of 100 grams. The thing you are trying to detect is smaller than the smallest number the instrument prints.
And how big is the noise?
Much bigger. A normal day-to-day swing from fluid, food volume and what has not yet left you is comfortably a kilogram, often two, and it does not need anything unusual to happen. Everybody knows this part; it is why the advice is always to weigh yourself under the same conditions each morning.
What gets less attention is the ratio. Signal: about 0.05 kg a day. Noise: about 1 kg a day. The noise is roughly twenty times the size of the thing you are looking for.
After four days of perfect adherence you have lost something like 0.2 kg of fat, sitting inside a band of variation five times that wide. A four-day stall is not weak evidence that nothing is happening. It is no evidence at all, and a four-day drop of 1.5 kg is equally meaningless in the other direction.
So how many days does it take?
Averaging is what rescues this. Average seven daily weigh-ins and the random part of the number shrinks by roughly the square root of the number of readings — around two and a half times for a week. A daily wobble of about a kilogram becomes a weekly average with maybe 0.3 to 0.4 kg of slop left in it.
Now compare two consecutive weekly averages. The signal between them is about 0.4 kg. The leftover uncertainty in the comparison is of a similar size, because both averages are noisy. Which means a single week-on-week comparison can easily come out flat, or slightly up, while you are losing fat steadily. Not rarely. Routinely.
Stretch the comparison to three weeks and the signal grows to well over a kilogram while the noise in each average stays where it was. Now the difference is two or three times larger than the uncertainty, and a flat line starts to mean something.
That is the whole answer: the signal accumulates with time, the noise does not. Waiting is not passivity, it is how the measurement gets made.
A practical version: compare the average of your last seven days against the average of the seven days ending three weeks earlier. If you are weighing yourself most mornings, that comparison is worth taking seriously. Anything shorter is a mood, not a measurement.
What does real progress actually look like?
Not a slope. Almost nobody gets a smooth descending line.
The usual pattern is stepwise: nine days of nothing, then 0.8 kg apparently overnight, then a week drifting slightly upwards, then another drop. Plotted daily it looks chaotic. Plotted as weekly averages it looks like a staircase going down. The flat sections are not stalls — they are the ordinary texture of the graph, and they are exactly the stretches during which people conclude the plan has failed.
The corollary is that a drop is not proof either. If you changed something six days ago and the scale fell, you have not learned that the change worked.
When the window really has been long enough, what do you check first?
Three or four weeks of averaged weigh-ins with no downward movement is worth acting on. Check in this order:
Intake drift, before anything else. It is far and away the most common cause, and it is rarely deliberate. Portions expand as food becomes familiar; the things that never got logged accumulate. A tablespoon of olive oil is about 120 kcal. A 30 g handful of nuts is about 180. Milk in four coffees is about 100. Two beers on a Friday are around 300. None of that feels like eating, and together it is a week's worth of deficit. Weigh and log everything for seven days — including weekends, which is where most of it hides — and compare it with what you assumed you were eating.
Then movement. Non-exercise activity quietly falls when you are eating less. Step counts from a month ago against step counts from this week will tell you in about ten seconds.
Then the target itself. Calorie targets from a formula are estimates built on population averages, and an individual can sit a few hundred kilocalories either side of what the formula says. A target inferred from your own data — what you logged set against what the scale did over a fortnight — is a better guess than a formula's, though still a guess. This is the approach Day So Far uses, and the honest framing is that it narrows the error rather than removing it.
Why changing things every few days is the real problem
Every adjustment resets the clock. If you cut 200 kcal on day five, change breakfast on day nine and add cardio on day twelve, you now have three overlapping experiments, none of them long enough to read, and a scale trend that cannot attribute anything to anything.
The frustrating truth is that impatience does not just fail to speed things up — it destroys the data you would need in order to find out what is actually wrong. Three weeks of an unchanged plan is worth more than three weeks of frantic optimisation, because only one of them produces an answer.
And when you do have your answer, the fix is not a bigger deficit. A deficit you cannot hold for a month is not a plan, it is a spike followed by a correction. If the evidence genuinely supports a change, make one modest change, write down the date, and hold it for another three weeks before you judge it.
Most stalls that people describe as stalls turn out to be somewhere between four and ten days long. Those ones do not need a fix. They need next Tuesday.