What to Eat After Poor Sleep: Adjusting Your Macros
How KEXBI dynamically recalibrates your macronutrient timing and targets the day after a short night, and the threshold where dietary interventions must stop.
~4 MIN · KEXBI RESEARCH
How KEXBI dynamically recalibrates your macronutrient timing and targets the day after a short night, and the threshold where dietary interventions must stop.
Four levers, calibrated to the evidence, with a hard floor where diet stops being the right intervention. How the KEXBI fuel plan responds the day after a short night.
The previous article (What Short Sleep Does) walked through the four mechanisms that a short night triggers in the body. This piece is about how those mechanisms translate into the actual numbers in your KEXBI fuel plan the next day.
Four levers do the work. Three are dietary; one is the engine telling you directly when diet is the wrong tool.
Breakfast carbs shift toward lunch and the pre-training window. Total daily carbs stay close to baseline.
When carbs do land at breakfast, the plan biases toward higher-fibre, lower-glycaemic options.
About +10 percent protein. The bigger lever for protecting muscle is intensity work, not extra shakes.
About 40 g casein, 30 minutes before bed, for overnight muscle protein synthesis.
Underneath the four levers is a floor. When your tracked sleep drops below about 5 hours for two consecutive nights, the macro adjustments stop scaling and the message changes. The honest framing is that diet is not the intervention at that point. The intervention is sleep.
The metabolic evidence in What Short Sleep Does points to a specific intervention: not a daily carb cut, but a timing shift. Peripheral insulin sensitivity is at its lowest in the morning after a short night, then partially recovers across the day [1,2]. The fuel plan responds by moving a portion of the usual breakfast carbohydrate into lunch and the pre-training window. Same total carbs. Better timed.
When carbs do land at breakfast on a low-sleep day, the engine biases toward higher-fibre, lower-glycaemic options. Same logic: smaller, slower glucose excursion through a temporarily slower disposal system. Oats with seeds and berries instead of toast and jam, beans on sourdough instead of pastries.
A note on what the plan does not do. Cutting daily carbs on a short-sleep day treats a problem the body is not actually presenting. Spaeth 2013 [3] showed the over-consumed macro after restriction was fat, not carbs, and Fang 2015 [4] showed the salience-network change correlated specifically with fat intake the following day. Cutting carbs would remove a macro that has genuine recovery value, particularly around training, and add a layer of restriction that erodes adherence. So the plan does not cut. It redistributes.
Peri-workout carbs are preserved. If you are training the same day, the carbs around your session do not move. The foundational performance evidence is robust enough that the plan defaults to leaving the peri-workout window alone. Coyle 1986 [5] showed in seven endurance-trained cyclists riding at 71 percent VO2max that carbohydrate feeding during exercise extended time to fatigue from 3.02 plus or minus 0.19 hours to 4.02 plus or minus 0.33 hours (P less than 0.01), maintaining euglycaemia and CHO oxidation when muscle glycogen was depleted. There is no sleep-specific RCT telling the engine to change peri-workout fuelling on a short night, and the case for preserving it is stronger than the case for guessing.
The catabolic profile after a short night is real (cortisol up, testosterone down, muscle protein synthesis down 18 percent per Lamon 2021). The instinctive response is to add protein. The evidence says training is the better lever.
Saner 2020 ran the experiment that resolves the question. Twenty-four young men, five nights at 4 hours in bed, randomised either to no exercise or to high-intensity interval cycling on days 2 through 4 [6]. Myofibrillar protein synthesis was significantly lower in the sleep-restricted, non-exercising group versus normal sleep. In the HIIE-during-restriction group, MPS was maintained at control levels. Training, not extra food, closed the gap.
So the day-after protein bump is capped at about 10 percent. Enough to give you a measurable target on a hard week without crowding out calories from other macros. Small enough that the plan is not pretending protein closes a gap the literature says training closes better. The accompanying coaching nudge is to keep training, especially the harder sessions, when sleep is the constraint.
On training days, the plan adds a pre-sleep casein nudge. About 40 g of casein, 30 minutes before bed, slow-release for overnight muscle protein synthesis. The protocol is anchored in two studies and a review.
Res 2012 took 16 young men through evening resistance work followed by 40 g of intrinsically-labelled casein 30 minutes before sleep [7]. Overnight muscle protein synthesis rates rose by approximately 22 percent versus placebo. Snijders 2015 extended the concept across 12 weeks of resistance training in 44 young men [8]: the protein group gained more strength (plus 164 plus or minus 11 kg combined 1RM versus plus 130 plus or minus 9 kg, P less than 0.001) and greater quadriceps cross-sectional area than the placebo group.
A small caveat. Snijders 2015 used 27.5 g of casein in its chronic protocol, not 40 g. The 40 g acute dose comes from Res 2012, and the convergence on 40 g as the dose that maximises overnight MPS in young adults is anchored in Trommelen and van Loon's 2016 review, which concludes that "at least 40 g of protein is required to display a robust increase in muscle protein synthesis rates throughout overnight sleep" [9]. The plan picks 40 g on the strength of that dose-response work. The chronic strength and mass benefits are extrapolated from the broader literature, not a single trial.
There is a point at which the fuel plan stops trying. When your tracked sleep is below about 5 hours for two consecutive nights, the day-after macro adjustments cap out. Carb timing nudges stop. Pre-sleep casein suggestions stop. The message shifts to a single line: sleep is the intervention.
The reasoning is Van Dongen 2003 [10]. In their statistical model, PVT lapses accumulated near-linearly with hours of wakefulness past a fitted critical threshold of 15.84 hours (SE 0.73 h). In the total-sleep-deprivation arm, by 36 to 48 hours awake, PVT lapse counts exceeded those produced by 14 days of 4-hour nights. Subjects did not notice. There is no realistic dietary lever that closes that gap. The most useful thing the plan can do at that point is refuse to pretend it can.
Three areas where the answer is not as clean as it should be, and the plan flags rather than pretends.
Morning workout collisions. If your training session is at 06:30 and you slept 4.5 hours, the "shift breakfast carbs later" rule collides with the "preserve peri-workout carbs" rule. There is no clean RCT that resolves this. The plan defaults to preserving the peri-workout window, but the magnitude on a short-sleep day specifically is a topic for a separate write-up.
Saturated-fat acute effects. Spaeth 2013 and Fang 2015 show the over-consumed macro after restriction is fat, and there is mechanistic work suggesting saturated fat specifically may worsen acute insulin resistance after a short night. The plan treats this as a hypothesis to test, not a rule to encode.
Wearable agreement. Different wearables disagree about your total sleep time by margins that occasionally matter (the 2025 meta-analysis covered in Not Eight Hours puts the average shortfall around 17 minutes versus polysomnography). The model handles this by smoothing across 14 days, but the underlying noise is real.
Day-after-short-sleep carbs are redistributed, not cut. Breakfast carbs shift later. GI drops at breakfast. Peri-workout carbs are preserved.
The protein bump on short-sleep days is capped at about 10 percent. The plan leans on training as the MPS lever, not on extra shakes.
On training days, you will see a pre-sleep casein nudge: about 40 g, about 30 minutes before bed.
Below about 5 hours of sleep for two consecutive nights, the fuel adjustments stop scaling. The app will tell you to sleep.
Open items are flagged in the app, not hidden. Morning-workout collisions, saturated-fat specifics, and the wearable-noise problem are areas where the model will iterate publicly as the literature moves.
The carb redistribution percentages in Fig 1, the 10 percent protein bump, the 40 g pre-sleep casein dose, and the 5-hour two-night threshold are KEXBI implementation values calibrated against the cited literature. They are not values reported by any single study. The Snijders 2015 protocol used 27.5 g casein; the 40 g acute dose is from Res 2012 and the dose-threshold argument from Trommelen 2016. The model will revisit these numbers as the field's evidence sharpens.