Deficit — The weekly check-in: what it is and why it works

Most calorie apps assume your TDEE is a fixed number. They calculate it once at signup, set your daily calorie goal, and then track you against it for as long as you stay subscribed. If the math is off by 150 kcal, you eat 150 kcal too much (or too little) every day, and the only signal you get is a scale that moves slower than you expected.

Real metabolism doesn't work that way. Your TDEE drifts as you lose weight, changes when you start training harder, and dips during a week of bad sleep. The day-one estimate stops describing you around week three.

Deficit handles this with a weekly check-in. Once every seven days, the app asks what your body did, you answer in about 60 seconds, and it adjusts your targets for the week ahead. Across an 8-week cycle, the targets converge on your actual metabolic response rather than what the formula predicted on day one.

What the check-in asks

On Sunday morning (or whichever day you started the cycle), the app shows a check-in flow with three or four screens, depending on how much data it already has from Apple Health and your daily logs.

Confirm your weekly average weight. The app computes the average of your morning weigh-ins from the last seven days. You verify it's right or adjust.
Optional: update body fat %. Once at the cycle midpoint and once at the end, the app asks for a fresh body fat reading. Tape measurements (Navy method), smart scale, or any method you have. Skipped weeks keep the previous reading.
Calorie reliability. A simple yes/no: were the calorie totals you logged this week accurate, or were there days you didn't track or guessed? This gates how aggressively the algorithm trusts the week's data.
Activity confirmation. The app already pulls steps and workouts from Apple Health. The check-in confirms whether your training and movement pattern matched the activity tier you set at cycle start, or whether it should re-bucket you.

The flow skips the diary entries, journaling, and interpretive charts that some apps include. It collects only the four signals the algorithm needs.

What the algorithm does with those answers

The check-in feeds a decision matrix. Each combination of (weight change × body fat change × calorie reliability) maps to a specific outcome. There are 12 outcomes for weight-loss cycles and 9 for maintenance cycles. The endpoints range from "perfect progress" to "stalled, possibly losing muscle, drop the deficit immediately."

A simplified slice:

Weight change	Body fat change	Calorie reliability	Action
On track (about −0.5 kg)	Trending down	Reliable	No change. Keep eating at current goal.
On track (about −0.5 kg)	Flat or up	Reliable	Possible muscle loss. Reduce deficit by about 10%.
Slower than expected	Any	Reliable	Real TDEE lower than estimate. Drop calorie goal about 5%.
Slower than expected	Any	Unreliable	No change. Re-collect cleaner data next week.
Faster than expected	Any	Reliable	Real TDEE higher. Raise calorie goal about 5%.
Up	Any	Reliable	Net surplus. Recheck deficit math; usually under-counting calories.
Anything	Any	Unreliable	Default to no-change. Algorithm doesn't trust noisy weeks.

The design rests on two principles.

The algorithm doesn't trust unreliable weeks. If you ate out four times and didn't log accurately, last week's "data" is too noisy to act on. Better to leave the goal alone and get a clean week before adjusting.
Adjustments are capped. No single check-in can change your calorie goal by more than ±10%. Even if the data strongly suggests a 15% adjustment is warranted, the cap forces convergence over several weeks rather than one overcorrection.

Why this beats fixed-number tracking

Fixed-number tracking has two failure modes: you eat too much, or you eat too little. Both are common because the day-one estimate is rarely exact.

Eating too much looks like: weight plateaus or rises, you blame compliance, you cut calories arbitrarily, you panic-cut harder, the deficit gets too steep, lean mass drops, body fat barely changes.

Eating too little looks like: weight drops fast in week 1 (water + glycogen), you assume the deficit is right, weight crashes 1.5 kg in week 2, you start losing strength, you can't sleep, training quality tanks, you abandon the cut at week 4.

The weekly check-in catches both. Eating too much shows up as "weight unchanged plus calories reliable," and the algorithm cuts the goal by about 5%. The real deficit emerges and things start moving again. Eating too little shows up as "weight dropping faster than expected plus body fat barely changing," and the algorithm raises calories. Lean mass loss stops and the cut becomes sustainable again.

Neither failure mode requires you to diagnose anything. The algorithm reads the data, decides what to do, and updates the goal. Your job is to log honestly and weigh in consistently. The rest runs by itself.

The first check-in is calibration, not judgment

Week 1 is special. You haven't eaten at the calculated goal long enough for the body to respond fully, so the data is noisy. Week 1 also includes the largest water-weight swings; glycogen depletion from the new diet can drop scale weight 1 to 2 kg in the first three days, which has nothing to do with fat loss.

Deficit's algorithm reflects this. Week 1 adjustments are softer than week 2+ adjustments. The system treats most of week 1's data as calibration noise, and only starts trusting weight-change signals by the end of week 2.

This is also why the trial period is 7 days rather than 3. Three days isn't long enough to cross the logging habit threshold or include even one weekly check-in. Seven days lands you right at the moment the system starts working. At minimum, you experience one data-driven adjustment before the trial ends.

What the check-in is not

A few things the ritual avoids by design.

Not a guilt trip. "Weight didn't move" is a data point, not a verdict. The algorithm trims the calorie goal and moves on.
Not a daily check-in. Daily weight is too noisy to trust. The data only stabilizes at weekly granularity.
Not a long flow. The four screens take under 60 seconds. Anything longer adds enough friction that users start skipping the check-in, and the algorithm runs blind.
Not optional in spirit. The app works without check-ins (your goal stays fixed at the day-one estimate), but that's the failure mode every other app has. Skipping the check-in turns Deficit into a worse version of MyFitnessPal.

What it adds up to over an 8-week cycle

By week 4, the algorithm has converged on your real TDEE within ±2%. Your calorie goal is honest, your body fat trend is downward, and the daily targets are tuned to the stress your body is carrying. The algorithm has corrected whatever Cunningham and PAL got wrong on day one.

By week 8, the cycle ends with two things you didn't have at the start: a measured body composition change, and a personalized TDEE number you can trust for future cycles. That second thing, the real maintenance number, is more valuable than the cut itself for everything you do afterward.

The weekly check-in is the smallest mechanism that gets you there. Four screens, 60 seconds, once a week. The math behind it does most of the work; the ritual itself does little.

Try for free