Run the number before you panic
The 3,500 calories per pound rule is a rough approximation. It overstates long term loss because it ignores the way energy expenditure drops as you get lighter, and I wouldn't use it to project six months out. As a same-day sanity check, though, it's perfect, because the numbers aren't close.
Say you tracked a 500 calorie deficit yesterday. Divide 500 by 3,500 and you get about 0.14 pounds of fat. Most bathroom scales read in 0.2 pound increments, so a full day of disciplined eating produces a fat change your scale can't even resolve. For the scale to honestly show 2 pounds of new fat, you would have needed a 7,000 calorie surplus in one day, which is a number you'd remember.
The same math works in reverse, and this is the part nobody wants to hear. A 2 pound overnight drop isn't 2 pounds of fat either. It's the same water leaving. Fat moves slowly in both directions. Everything fast on the scale is fluid, glycogen, or gut contents.
What the two pounds is made of
Five things account for nearly all short term scale movement.
- Glycogen and the water bound to it. Your body stores carbohydrate as glycogen in muscle and liver, and each gram of glycogen holds roughly 3 grams of water with it. Full stores in an average adult run somewhere in the range of 400 to 600 grams, which is several pounds of scale weight once the bound water is counted. Eat a big pasta dinner after a low carb stretch and you refill that overnight.
- Sodium. Sodium pulls water into the space outside your cells until your kidneys rebalance, which typically takes a day or two. Soy sauce, deli meat, canned soup, most restaurant food. The swing isn't proportional to your total sodium so much as to the change from what your body has been used to.
- Food and drink in transit. A 32 ounce water bottle weighs 2 pounds the second you drink it. Food doesn't evaporate when you swallow it. A high volume, high fiber day leaves more mass in your gut at any given moment, and a day without a bowel movement leaves more still.
- Training and inflammation. A hard leg session, or the first week of any new program, causes muscle damage and a fluid response around it. If you started creatine, it loads water inside the muscle cell and commonly adds a couple of pounds in the first few weeks that tends to stay while you keep taking it.
- Hormonal cycling. People who track through a full menstrual cycle usually see a rise in the days before their period that clears within a few days of it starting. If you only compare cycle day 26 to cycle day 5, you'll scare yourself for no reason. Compare the same phase to the same phase.
Alcohol, poor sleep, a long flight, a stressful week, and being sick all push in the same direction. None of them are fat.
The seven day average, worked out
Here are two weeks of morning weigh-ins from someone eating in a modest deficit. These are made up numbers, but the pattern is the one I see in my own data constantly.
| Day | Week 1 (lbs) | Week 2 (lbs) |
|---|---|---|
| Monday | 184.2 | 184.0 |
| Tuesday | 183.8 | 183.2 |
| Wednesday | 184.6 | 183.6 |
| Thursday | 183.4 | 182.8 |
| Friday | 183.9 | 183.5 |
| Saturday | 185.8 | 185.0 |
| Sunday | 185.1 | 184.3 |
| 7-day average | 184.4 | 183.8 |
Look at Saturday. Both weeks spike, because Friday night is when the sodium and the extra carbs land. If this person had weighed on Thursday of week 1 (183.4) and then again on Saturday of week 2 (185.0), they'd have concluded they gained 1.6 pounds in nine days and cut their calories out of frustration.
The averages tell a different story. 1290.8 divided by 7 is 184.4. 1286.4 divided by 7 is 183.8. That is about 0.6 pounds down week over week. Multiply 0.6 by 3,500 and divide by 7 and you land near a 300 calorie daily deficit, which is a reasonable, unglamorous rate of change and matches roughly what the person was aiming for.
A plain seven day mean is the version you can do in your head. Some people prefer an exponentially weighted average, the approach John Walker popularized in The Hacker's Diet, which gives recent days more pull and reacts a little faster. Either works. Consistency beats sophistication here.
How to weigh so the average means something
The average is only as good as the inputs, and the inputs are easy to corrupt without noticing.
- Weigh first thing in the morning, after using the bathroom, before food or drink. This is the most standardized your body gets all day.
- Same clothing situation every time. Naked, or the same pair of shorts.
- Same scale, same hard floor, same spot. Carpet compresses under the feet of the scale and reads wrong.
- Weigh every day if you can handle it. Seven data points make a much steadier average than three. If daily weighing wrecks your head, pick three fixed days, average those, and accept a noisier signal.
- Log the high ones. This is the rule people break. Skipping the day after a birthday dinner biases your average downward and hides the trend you're trying to read.
- Never compare a single day to a single day. Compare this week's average to last week's average.
Situations where the scale gets extra noisy
GLP-1 medications
If you're on a GLP-1, your dosing, your titration schedule, and any change to either are between you and your prescriber. Nothing here is medical advice, and I am a developer, not a clinician. On the scale specifically: intake often drops fast in the first weeks, which means less food and fluid moving through you at any moment and a real change in glycogen stores, so early scale movement isn't all fat, and it usually doesn't continue at the same rate. If you can't keep fluids down or the scale is moving in a way that worries you, that's a phone call to your clinician.
Starting or ending a low carb phase
The dramatic first week of keto is mostly glycogen water leaving. Reintroducing carbs puts a chunk of it straight back. Neither direction is fat, and both fool people badly.
Starting creatine, or a new training block
Expect an upward step that then holds. Reset your expectations to the new baseline instead of reading it as failure.
When to change something, and when the scale is the wrong instrument
Give any change two to three weeks of trend data before you touch your calories. One week isn't enough to separate signal from noise for most people, since day to day noise of a pound or more can swamp a week of real progress.
If your seven day average is flat or climbing across three straight weeks, the first place to look is intake accuracy. Weekend meals that never got logged, oil in the pan, a splash of cream in four coffees, portions estimated by eye instead of weighed. Forty grams of chicken misjudged is nothing; a tablespoon of olive oil unlogged twice a day is roughly 240 calories, and that alone can cancel a deficit.
Also accept what a bathroom scale can and can't see. It measures total mass, so it can't tell you what fraction is fat, muscle, water, or last night's dinner. If you're new to lifting or eating high protein in a deficit, body composition can change while the number sits still for weeks. Waist circumference measured at the same spot, how a specific pair of pants fits, and what you can lift are all worth tracking next to the weight. Losing weight isn't linear for anyone, and flat stretches are normal.