How to choose a calorie deficit: the science of adherence

"What are you working towards?"

That question lands on the third screen of a five-phase setup. What most people miss is that the screens before it, and the screens after, matter just as much. The phases are Welcome, Foundation, Reveal, Goals, and Context and Ignition. Fifteen minutes, if you move at a normal pace. The quality of what you put in is the ceiling on what any coach or plan can do for you.

This piece walks each phase and names the choice that matters most. Where research sharpens a call, it is cited. Where a number is illustrative, it is flagged.

Chapter 01 · WelcomeStart honest.

The Welcome phase asks for your name and your consent to a short set of terms. That is all it asks. It looks like throat-clearing. It is not.

The only choice that matters here is a disposition. You either treat the next twenty-five screens as data entry, or you treat them as a contract you are writing with yourself. People who cut corners in Welcome cut corners in Foundation. People who cut corners in Foundation get a plan that fits a person who does not exist.

Chapter 02 · FoundationThe body data.

Foundation asks for the inputs that shape everything that follows: height, weight, sex, date of birth, body fat percentage, and a picture of how you actually move through your week. Activity level. Training style. Training volume.

The single highest-leverage input is body fat percentage. Most weight-tracking apps skip it and work from weight alone. KEXBI reads lean mass directly when body fat is present, rather than inferring it from population averages. A person at 85 kg with 75 kg of lean mass burns more at rest than a person at 85 kg with 64 kg of lean mass. The engine knows which one you are only if you tell it.

From most accurate to least: DEXA, BIA smart scale, caliper, visual body fat chart. All four beat a blank field. If you cannot get a DEXA this week, a BIA reading or a visual estimate is still a real input. You can refine it later. What you cannot do is undo twelve weeks of a plan that assumed you were ten percent leaner than you are.

The two Foundation fields most people fudge are activity level and training style.

Activity level is about your day job and general movement, not your training. A desk worker who runs three times a week picks Sedentary. The runs get logged separately as training. Picking Moderately Active because you "feel active" double-counts the runs and inflates the target calories by roughly ten to fifteen percent, which in Fat Loss mode can be the difference between losing and not.

Training style is the same principle. Pick what your training looks like across a month, not what it looks like on your best week. If you run five times a week and lift twice, you are Endurance, not Strength and Power. The choice sets a protein floor appropriate to the stimulus you are actually putting the body under.

Chapter 03 · RevealReading your numbers.

The Reveal phase is one screen. It shows your baseline: BMR, TDEE, lean body mass, baseline hydration. Two things to know about these numbers.

First, they are calculated from the inputs you just entered. If a number looks wrong, the input was wrong. Go back.

Second, do not overfit to them on day one. The TDEE on the reveal card is a best first guess from equations. By day eight, if you have been logging, KEXBI starts re-learning your TDEE from what your body is actually doing: how much you ate, how your weight moved, how your training landed. The number on the reveal card today is not the number that runs your plan in six weeks.

This matters because new users often look at the TDEE on the reveal, compare it to what they eat now, and immediately want to eat less or eat more. The right move is to eat at the suggested target for a week, log accurately, and let the engine triangulate.

Chapter 04 · GoalsWhat each one means and who it is for.

This is where most plans live or die. One question, five tiles.

Fat Loss is for someone whose current body fat is above where they want it, and who is willing to sit in a caloric deficit long enough for the body to move. Right for someone with real fat to lose (typically above 15 percent in men, above 22 percent in women). Wrong for someone already at 10 percent who is convinced they need to cut for three pounds.

Maintain is the most underused tile. It is the right call after a cut, during a physically demanding life stage, or when you have been chasing targets for two years and need to stabilise. Maintenance is not a step backwards. It is the phase where your weight can settle.

Lean Gain is for someone already close to their ideal body fat who wants to add muscle. It requires a small caloric surplus and training volume that can actually use it. Picking Lean Gain without the training to back it is likely to turn more of the surplus into fat.

Recomp is for the narrow band of people who are relatively lean (under about 18 percent in men, 25 percent in women) but are new to serious training, or returning after a long break. Recomp asks for a small deficit and elevated protein, and the novice-training stimulus drives muscle gain while body fat falls. For experienced trained adults, recomp is slow and often disappointing.

Performance is for people in or near a competitive block. The calorie bias is neutral to slight surplus. Protein is elevated. This is the wrong Goal for someone who wants to look different. It is the right Goal for someone who wants to run, lift, or ride further.

Once the Goal is set, the next screen is the pace slider. This is the lever people overturn.

For Fat Loss, the slider runs from roughly one percent deficit to twenty-five percent. The default is ten. On average, people drag it harder than that. They should not.

Research suggests smaller deficits are easier to sustain. Continuous deficits of fifteen percent or less look sustainable across twelve to sixteen weeks for most people [2,3]. Above twenty percent, dropout tends to rise. At twenty-five percent, a trained adult is relying on behaviour that very few people sustain outside a short contest prep.

Rate of loss follows the same curve. Research on trained people points to 0.5 to 1 percent of bodyweight per week for weight loss that keeps lean mass [4]. Faster loss tends to take more of it from lean tissue. The scale moves faster. The body composition can be worse.

KEXBI uses the Energy Availability band underneath the slider as a planning guardrail. Research treats low energy availability as a continuum, not one universal cut-off [5,6]. KEXBI therefore works to keep the calories left after training above about 30 kcal per kg of lean mass for women and 25 for men, for people training at a moderate level or above. Below that, the band turns amber as a plain heads-up and the Save button disables. Read it as KEXBI declining to write a plan that under-fuels your training.

One more note. On any Fat Loss plan longer than twelve weeks, KEXBI inserts a two-week maintenance interval at full TDEE at weeks thirteen and fourteen, and every fourteen weeks after. The break is in the plan before you start. Research suggests intermittent diet breaks do not meaningfully protect metabolism over a continuous deficit in trained adults, but may help adherence [7,8]. KEXBI therefore uses two weeks at maintenance as a psychologically recoverable stop.

Chapter 05 · Context + IgnitionSetting up for adherence.

The last phase looks like admin. It is the part of the setup that most decides whether you finish.

Dietary preference tells the meal engine which cuisines and patterns to work from. Pick your real default kitchen. Flexibility if you eat most things. Mediterranean if that is where most of your cooking already sits. Keto only if you have actually run keto before. Aspirational cuisines produce meals you will not cook.

Pantry is where most users move the needle without noticing. Mark the staples you actually have. Rice, oats, eggs, olive oil, canned beans. KEXBI's meal generation prefers ingredients you own. The closer the pantry reflects your kitchen, the less friction every meal has.

Favourite activities set the shape of your week in the training calendar. Pick activities already in your life, not activities you want in your life but have not started. You can add more later. Friction in week one makes drop-off by week three more likely.

Sleep schedule is not a preference. It anchors the circadian layer of your daily plan. If you sleep from eleven to six, enter that. Not "I want to go to bed at ten." The engine times protein distribution, training windows, and evening meal composition around the schedule you give it. A lie here shows up as meals served at the wrong hour and a plan that feels out of sync. Sleep quality has its own downstream effect: short sleep can shift hunger hormones enough to make a clean deficit feel like a deep one [9,10].

Weigh-in frequency matters more than most users expect. KEXBI advocates daily, and the evidence points the same way. In a systematic review, more frequent self-weighing was linked to more weight lost and better maintenance [12]. The 2014 Steinberg RCT on psychological safety found no increase in depression, disordered eating, or body dissatisfaction from daily weighing in adults [13], which runs against the intuition that seeing the number every day breeds anxiety.

The rationale is signal, not willpower. A single daily reading is noisy: water, glycogen, yesterday's sodium, where you are in your cycle. A seven-day rolling average is clean. Daily data produces a tighter trend line than weekly data, which means the engine can catch a stalled week in four days, not in fourteen. Compressed feedback loops help separate plans that finish from plans that drift.

One caveat. If you have a history of disordered eating, or the number on the scale drives unhealthy behaviour rather than informing your plan, pick weekly. Understand the trade-off. Fewer data points means a slower response, and the engine holds adjustments longer while it waits for signal.

HealthKit and notifications are the final gate. Without HealthKit, the engine runs from the answers you gave in Foundation and the meals you log. With HealthKit, the engine is watching the day: sleep, steps, heart rate, workouts. If you are taking KEXBI seriously, grant it. The difference in plan quality is not small.

Chapter 06 · The takeawayStart well, train well.

Onboarding is not a questionnaire. It is the contract you sign with your own plan. Honest inputs produce a plan that fits your life. Approximated inputs produce a plan that fits a composite fiction, and the next twelve weeks feel wrong without being obviously wrong.

Spend the fifteen minutes. Measure the body fat if you can. Pick the Goal that matches today, not the one that flatters you. Set the slider to the range you can finish from. Fill in the pantry and the sleep schedule like they matter, because they do.

Start well and the rest of the work is training and eating.