How to Find True Maintenance Calories: Why Trackers Fail
Your metabolic rate is a dynamic feedback loop, not a static spreadsheet calculation; holding true maintenance requires constant, data-driven correction.
~4 MIN · KEXBI RESEARCH
Your metabolic rate is a dynamic feedback loop, not a static spreadsheet calculation; holding true maintenance requires constant, data-driven correction.
Ask a lean, trained person which of the four goal states is hardest to hold and they will name maintenance every time. Cutting has a clear direction and a clear stop line. Bulking has a clear direction and a scale trend to correct against. Maintenance is a moving target dressed up as a single number, and the app on your phone is the wrong tool to hold it.
The textbook picture of energy expenditure is four boxes: basal metabolic rate, the thermic effect of food, the energy cost of exercise, and non-exercise activity thermogenesis. Add them up and you get TDEE. The picture is useful for teaching. It is misleading for planning maintenance, because it implies the four numbers are stable inputs.
They are not. The body defends its weight actively. Rosenbaum and Leibel 2010 summarised the evidence for adaptive thermogenesis: after weight loss, measured resting metabolic rate drops more than the fall in tissue mass predicts, and the reduction persists [1]. The Biggest Loser follow-up, Fothergill and colleagues 2016, tracked contestants six years after a large weight loss and found average resting metabolic rate remained about 500 kcal per day lower than expected for their current weight [2]. Adaptive thermogenesis is real and durable.
NEAT is the noisiest term in the equation. Levine and colleagues showed that within-individual NEAT changes with intake (overfeeding raises it in some people, restricted feeding lowers it) and that between-individual variation in NEAT reaches roughly 2,000 kcal per day among free-living adults [3]. Two people of the same weight and lean mass can differ by that much on a given day, and the same person can differ by hundreds of kilocalories week to week.
Add these two effects together and the "TDEE number" you were handed at the start of your maintenance phase is out of date within weeks.
A standard tracker asks for your age, sex, height, weight, and activity level, hands back a calorie target, and calls the setup finished. The target is a Mifflin-St Jeor or Harris-Benedict estimate multiplied by a fixed activity factor. Neither of those inputs updates itself in response to your last twelve weeks of training or your last diet phase. The app records what you ate against a number it computed on day one and treats "hitting it" as success.
Two problems follow.
The maintenance number is wrong on day one, for most people, in one direction or the other. Mifflin-St Jeor over-estimates measured RMR by 5 to 10 percent in a large fraction of adults [4], and Harris-Benedict is worse. The activity-factor multiplier is a categorical guess. The person who reports "moderately active" and the person who reports "very active" may occupy the same real position; the person who logs one training session a day may still have half the NEAT of the person who does not sit still. The starting estimate is bound to be off by a few hundred kilocalories in either direction.
The maintenance number then drifts, and the app cannot see it. Metabolic adaptation from a prior cut suppresses RMR; a hidden bulk raises it; illness or a busy week at work drops NEAT; a new training block raises it. Weight-stable at one intake does not mean weight-stable at the same intake next month. Body composition can shift silently while the scale reads the same number, because a kg of lean tissue and a kg of fat tissue occupy the same 1 kg on the scale but do very different work on TDEE.
The person on the left did their job. They hit the tracker's number every day. The tracker cannot tell them they were slightly under maintenance for their lean tissue and slightly over for their fat tissue, so they traded a small amount of muscle for a small amount of fat while the scale held steady.
For an athlete, the point of a maintenance phase is not to keep a specific number on the scale. It is to preserve the body composition you built, defend the training you can do in that body, and let adaptive thermogenesis settle after a cut or a bulk. The right maintenance number is the one that holds those three things at once.
Three levers matter more than the daily calorie total.
Protein intake. Higher protein defends lean mass in energy balance and in a deficit, with meaningful benefits up to about 1.6 to 2.4 g per kg body weight per day [5,6]. In maintenance, this range keeps synthesis primed even on non-training days.
Training stimulus. Resistance training raises muscle protein synthesis for 24 to 48 hours after a session. During maintenance, keeping a resistance-training stimulus on the calendar preserves the lean mass that maintenance is trying to hold [7].
Session-day energy. Days with a hard training session need more fuel than days without one. Handing the athlete a single flat number across the whole week averages the two, and the average is wrong on both days. Training days come in under; rest days come in over.
The period after a fat-loss phase is where standard trackers fail most predictably. The athlete arrives at goal weight with a suppressed RMR, an elevated hunger drive, and a metabolism that is not yet at the level the tracker's estimate assumes.
The correct move is a controlled reintroduction of calories: raise intake in small weekly increments and track body composition, hunger, and training output as the metabolism recovers [8]. What a fixed-target tracker does instead is snap the target from deficit-mode to a computed "maintenance" that is a few hundred kilocalories too high, because the computation cannot see the adaptation. The predictable outcome is fat rebound in the first few weeks, followed by the athlete believing they are "the kind of person who gains weight easily."
Maintenance is not a single-number problem. It is a control-loop problem: measure, correct, measure again. KEXBI runs the loop for you, and the mechanism has three inputs it wants you to keep feeding.
The planned intake is what the engine put on the plate for the day. The tracked intake is what you actually logged as eaten. The weigh-in is what the scale returned. Given all three across a rolling window, the engine back-solves your actual TDEE from the numbers themselves: measured intake minus the energy stored or released by your observed weight change equals what your body must have burned. Nothing predictive, nothing borrowed from a population equation; the number your body produced this week, from what you actually did.
The tracker was built to hit a fixed calorie target. Maintenance is not a fixed calorie target. It is what your body settles on this week, given three months of adaptation, a training week that is different from last week's, and NEAT that moves with your intake and your sleep. Holding it takes measurement, correction, and a target that updates. The right answer is not a smarter starting number. It is a control loop that runs every week, holds body composition rather than body weight, and treats the scale as a signal rather than a scoreboard.