In this LAB note
Question · Hypothesis · Method · Results · What Failed · What We Learned · What Changed at MOMS
Question
Every one of our 201 recipes lands at or above 30.0 g of protein and at or below 399.9 kcal. Neither line is ever crossed. Which of those two walls is actually binding on the formulation — and which ingredient slot pays for the recipes that sit hardest against it?
Hypothesis
We expected the calorie ceiling to be the constraint that hurts. A 16 oz whole-food shake has to carry a protein source, a liquid, a produce load and a texture ingredient inside roughly 400 calories, and calories are the thing every one of those slots spends. Protein, by contrast, we expected to be the easy target: whole-food protein sources are dense enough that once you have committed to one, 30 g should arrive with room to spare, and the catalogue should spread out well above the floor. We also expected slack to be productive — that a recipe finishing well under 400 calories would spend the space it saved on more produce, more fiber, or a functional ingredient.
Method
Two published results set the floor we were testing against. Mamerow and colleagues (Journal of Nutrition, 2014) ran a seven-day crossover feeding study in eight healthy adults, comparing an isoenergetic, isonitrogenous diet delivering roughly 30 g of protein at each of three meals against the same daily total skewed toward dinner (about 10 g at breakfast, 15 g at lunch, 65 g at dinner). Moore and colleagues (Journals of Gerontology Series A, 2015) fed graded protein doses to healthy younger and older men and identified the dose at which myofibrillar protein synthesis stopped rising. Internally, we took all 201 recipes in the MOMS formulation database — every one specified at a 16 oz serving — and pulled Kcal and Protein (g) for each, then plotted both against their limits. We binned recipes by distance from the 30 g protein floor and by distance from the 400 kcal ceiling. We then parsed each recipe’s six ingredient slots (Liquid, Body, Fruit & Vegetable, Protein, Functional, Flavor) into individual ingredients with their gram weights, and compared slot composition between the recipes pressed hardest against the floor and the rest of the catalogue.
Results
Both walls hold exactly, and neither is ever crossed. The lowest-protein recipe in the database is MY-W2-DETOX-009-V1 ("Dewdrop Detox") at exactly 30.0 g. The highest-calorie recipe is WY-W1-DETOX-017-V1 ("Amber Detox") at 399.9 kcal. Nothing sits at 29 g, and nothing sits at 401 kcal. But the two walls are not equally crowded. Calories have room: the range runs from 270.5 to 399.9 kcal, a 129 kcal spread, and only 40 of 201 recipes come within 5 kcal of the ceiling. Protein does not. The median recipe carries 32.4 g, clearing the floor by 2.4 g, and 38 recipes clear it by less than a single gram. Seventy-five of 201 — more than a third of the catalogue — finish under 32 g. Against a maximum of 48.2 g, the distribution is not centred in its range. It is stacked against the low wall. The recipes in that corner are recognisable. EI-W3-GUT-005-V1 ("Gentle Gut", Standard) finishes at 390.0 kcal and 30.8 g of protein. Its entire protein slot is Egg White 240 g. It carries a heavy produce load — Pineapple 91 g and Kiwi 81.2 g, on top of Banana 59.1 g and Avocado 31.3 g in the body slot — and both its Functional and Flavor slots are empty. EY-W1-ENERGY-008-V1 ("Golden Charge", Meal Replacement) is the same shape reached from the other direction: 397.4 kcal, 30.5 g protein, and a protein slot of four ingredients totalling 245 g (Chickpeas 60 g + Soy Milk 120 ml + Black Beans 35 g + Egg White 150 g). Its Functional slot is empty too. Twelve recipes sit within a gram of the protein floor and within 10 kcal of the ceiling at the same time. Across the 38 floor recipes as a group the pattern is consistent. They carry more produce than average — 104.1 g in the Fruit & Vegetable slot against 92.3 g across all 201 — and more carbohydrate (41.7 g against 39.5 g). What they do not carry is a functional ingredient. Only 42 percent of the floor recipes have anything in the Functional slot, against 62 percent of the catalogue as a whole. The slot that gets dropped is the smallest one in the recipe: across all 201 recipes the Functional slot averages 4.65 g of food. So the protein floor is the binding wall, not the calorie ceiling — and the thing it displaces is not produce. It is the five grams of chia, flaxseed, cashew or turmeric the recipe would otherwise have carried.
What Failed
Three things, and the third one constrains what this note is allowed to claim. First, our slack hypothesis is wrong in the plainest way. EY-W2-DETOX-006-V1 ("Airy Cleanse") finishes at 270.5 kcal — 129 kcal below the ceiling, the most headroom of any recipe in the database — and spends none of it. Its entire produce slot is Carrot 50 g. It finishes with 3.2 g of fiber, less than half the catalogue mean of 7.6 g, and its Functional slot is empty as well. Calorie room is not automatically converted into food. Here it is simply left unspent, and the recipe is thinner for it. Two of the three lowest-calorie recipes in the database look the same way. Second, our Score does not see any of this. Score correlates with protein at r = +0.070 and with calories at r = +0.022 — both indistinguishable from zero. A recipe wedged into both corners and a recipe with 129 kcal of slack get the same treatment from our internal quality number. We would like to say the Score rewards tight formulation. It does not measure it, so we cannot say that, and nothing here should be read as evidence that the constrained recipes are the better ones. They are only the more constrained ones. Third, and most important: this data cannot be used to argue that 30 and 400 are the right numbers. A distribution that runs right up to two round figures and stops dead at both is the signature of a rule imposed before the recipes were written, not a boundary the food found on its own. The published work supports a defensible floor — Mamerow and colleagues measured 24-hour mixed muscle protein fractional synthesis rate 25 percent higher on the even 30 g-per-meal distribution than on the skewed one, and note that meals below 30 g attenuate the postprandial response — but that study had eight participants. And Moore and colleagues make the more awkward point: the plateau dose is relative to body mass, at 0.24 ± 0.06 g/kg in younger men and 0.40 ± 0.19 g/kg in older men. For a 55 kg woman, 30 g sits well past the younger-adult plateau; for a 90 kg older adult it sits under the older-adult one. Our floor is a single fixed number applied to every body type, inside a product built entirely on the premise that body types differ. That is an inconsistency in our own system. This analysis surfaced it. It did not resolve it.
What We Learned
The box our recipes live in is not symmetric, and we had the tight side wrong. We had been managing calories, because calories are the number customers ask about and the number a formulator watches while adding ingredients. The wall the recipes are actually pressed against is the protein floor — and it is pressed hardest by the recipes that reach 30 g through high-volume, low-density protein sources, whether that is 240 g of egg white or four ingredients totalling 245 g. Those sources take up the room in the blender that the small ingredients needed. The displacement is also quiet, which is why we had not seen it. Losing five grams of chia or flaxseed changes almost nothing on a nutrition panel. It is a rounding error in calories and close to a rounding error in fiber. It is also the entire functional slot.
What Changed at MOMS
We now record two headroom figures on every recipe rather than only the finished macros: calories below 400, and grams of protein above 30. Both are formulation fields now, not outputs read off at the end. Any recipe finishing with more than 60 kcal of headroom is flagged, and has to either spend it — on produce, fiber, or a functional ingredient — or be reclassified to a lighter Cal Label rather than passed as-is. Any recipe within 1 g of the protein floor is now reviewed specifically for whether its Functional slot was emptied out of necessity or out of habit. Where it is habit, the protein slot gets rebuilt around a denser source before the functional ingredient is sacrificed. We have also opened the harder question the Moore data raises: whether a flat 30 g floor belongs in a system that already personalises by body type, or whether the floor should scale with body mass the way the published dose-response does. We are not changing it yet. We have written it down as an open contradiction rather than leaving it implicit.
References
Mamerow MM, Mettler JA, English KL, Casperson SL, Arentson-Lantz E, Sheffield-Moore M, Layman DK, Paddon-Jones D. Dietary Protein Distribution Positively Influences 24-h Muscle Protein Synthesis in Healthy Adults. Journal of Nutrition, 144(6):876-880, 2014. | Moore DR, Churchward-Venne TA, Witard O, Breen L, Burd NA, Tipton KD, Phillips SM. Protein Ingestion to Stimulate Myofibrillar Protein Synthesis Requires Greater Relative Protein Intakes in Healthy Older Versus Younger Men. Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 70(1):57-62, 2015.
Keep reading
Previous experiment
Not sure which shake fits you? We match the formulation to your body type.
Find your body type