What the scale actually measures

Not fat. The scale measures the total mass standing on it: fat, muscle, bone, water, glycogen, this morning’s coffee, and yesterday’s dinner still in transit.

Fat changes slowly. Everything else on that list swings by the day:

Water. ±1 kg between days is unremarkable. A salty dinner alone can hold an extra liter.
Glycogen. Each gram of stored carbohydrate binds roughly 3 g of water. Refilling depleted stores after a high-carb day adds 1–2 kg overnight.
Gut contents. A high-fiber day vs a low-fiber day is easily 0.5 kg of difference in transit.
Cycle phase. For many women, water retention adds 1–2 kg in the late luteal phase and releases within days.

None of this is fat. All of it shows up on the scale with the same authority.

The signal is 30x smaller than the noise

Here’s the core problem as arithmetic. At a 500 kcal daily deficit — a solid, deliberate cut — you lose about 65 g of fat per day (7,700 kcal per kg of fat, 500/7,700 × 1,000 ≈ 65).

The daily noise is ±1,000–2,000 g. You’re trying to read a 65 g signal through noise 20–30 times larger. A single weigh-in cannot tell you whether the diet is working. Two weigh-ins a week apart barely can. This isn’t a discipline problem or a metabolism mystery — it’s signal processing.

The fix isn’t weighing less. It’s weighing more and reading the data differently.

How to weigh so the data is usable

Daily, under fixed conditions, so the noise at least stays consistent:

Same time — morning works best.
After the bathroom, before food or water.
Same scale, same spot on a hard floor (carpet skews readings).
Log the number and move on. No reaction, no interpretation. One data point.

Ten seconds a day. The number itself is allowed to be ugly — you’re not reading it yet.

How to read the trend

Once a week, compare this week’s average to last week’s average. Averages cancel most of the water noise; single points don’t.

What a sane cut looks like: 0.25–0.75% of body weight per week. For an 80 kg person that’s 200–600 g per week — which, note, is smaller than one day of noise. Hence the averaging.

In Calsheet, the weight journal in Stats shows your entries over 7 or 30 days plus the net change over the period. Read the shape of the 30-day trend and ignore individual points. A jagged line drifting down is success. A smooth line was never on offer.

Three spikes that fool everyone

The Monday spike. After a weekend of eating out, the scale jumps +1.5 kg. To gain 1.5 kg of fat you’d need roughly 11,500 kcal over maintenance in two days. You didn’t do that. It’s glycogen plus water plus salt, and it drains within 2–3 days. Keep logging, don’t compensate with a starvation Monday.

The salt spike. Ramen, pizza, anything restaurant-salty: +1 kg next morning, gone by Thursday. Same mechanism, shorter timeline.

The new-program masking. Started lifting? Muscles under new load retain water for repair — often 1–2 weeks of scale stall while fat loss continues underneath. The trend catches up. Quitting during the mask week is the classic own-goal.

When to actually change your calories

Only when all three hold:

The weekly averages have been flat (or moving the wrong way) for 2–3 weeks — not days.
Logging has been honest the whole time, weekends included. Unlogged weekends are the most common “mysterious plateau.”
No masking event in play (new training program, cycle phase, travel and salt).

If all three are true, adjust by a small step — 100–150 kcal off the daily budget — and give the new number another 2–3 weeks. In Calsheet the daily goal is editable in Settings; the wizard’s estimate is a starting point, not a verdict. TDEE formulas carry ±10% error by design, and your logged reality outranks any formula.

If daily weighing messes with your head

Real trade-off, worth stating plainly: for some people a red number at 7 a.m. taxes the whole day, and for anyone with a history of disordered eating, daily weighing can do more harm than the data is worth.

The fallback: weigh once a week, same fixed conditions, and accept that the trend needs 4–6 weeks to become readable instead of 2–3. Slower feedback, intact sanity — a fair trade. The scale is also just one instrument: a tape measure at the waist and how your jeans fit are slower but harder to fool.

TL;DR

The scale measures fat plus 1–2 kg of daily water/glycogen/salt noise. Real fat loss is ~65 g/day — 30x smaller than the noise.
Weigh daily under fixed conditions, log it, react to nothing.
Judge by weekly averages and the 30-day shape. Sane rate: 0.25–0.75% of body weight per week.
Monday spikes, salt spikes, and new-gym-program stalls are water. They resolve in days; don’t compensate.
Change your budget only after 2–3 weeks of flat weekly averages with honest logging — then by 100–150 kcal, not 500.