In this LAB note
Question · Hypothesis · Method · Results · What Failed · What We Learned · What Changed at MOMS
Question
Why does a protein slot weighing 365 grams deliver less protein than one weighing 60 grams?
Hypothesis
We expected protein-slot mass to work as a rough proxy for protein delivered. When a formulator sketches a recipe, the working shorthand is volume: a bigger scoop of the protein ingredient should mean more protein in the finished shake. We predicted a moderate positive correlation across the database — not perfect, since ingredients differ, but clearly positive. We also expected the exceptions to be rare, confined to a handful of unusual recipes rather than describing the general case.
Method
We parsed all 201 recipes in the MOMS formulation database and extracted every gram and milliliter weight listed in the Protein column, summing them into a single slot mass per recipe. We then compared that mass against the recipe's measured Protein (g) and computed a yield figure for each shake: grams of protein delivered per gram of food placed in the protein slot. We ran correlations of slot mass against protein delivered, fiber, and calories, and grouped the database by dominant protein ingredient and by body type. We read that against two published sources. Berrazaga and colleagues (Nutrients, 2019) reviewed plant- versus animal-based protein for muscle mass maintenance and concluded that plant proteins have a weaker anabolic effect owing to lower digestibility and lower essential amino acid content, and that increasing protein intake produced a positive acute postprandial muscle protein synthesis response, with blending several plant protein sources listed among the strategies for improving amino acid composition. Holt and colleagues (European Journal of Clinical Nutrition, 1995) fed isoenergetic 240 kcal servings of 38 foods to groups of 11-13 subjects and found satiety index scores correlated positively with the serving weight of the food (r = 0.66, P < 0.001, n = 38) and with its water content (r = 0.64, P < 0.001), and negatively with fat content (r = -0.43, P < 0.01).
Results
Protein-slot mass across the 201 recipes ranges from 60 g to 365 g — a 6.1-fold spread. Protein delivered ranges from 30.0 g to 48.2 g, a 1.6-fold spread. The correlation between the two is not merely weak. It is negative: r = -0.22. Heavier protein slots deliver slightly less protein, not more. The two endpoints sit almost exactly on top of each other in every other respect. WAI-W1-WARM-010-V1 "Golden Warmth" carries Chicken 60 g and delivers 32.0 g of protein at 394.6 kcal. EY-W1-ENERGY-008-V1 "Golden Charge" carries Chickpeas 60 g + Soy Milk 120 ml + Black Beans 35 g + Egg White 150 g — 365 g of food — and delivers 30.5 g of protein at 397.4 kcal. That is 305 additional grams of food, 1.5 grams less protein, and a 2.8 kcal difference between the two shakes. Yield makes the mechanism explicit. It ranges from 0.084 to 0.533 g of protein per gram of slot food, a 6.3-fold spread that almost exactly mirrors the mass spread. Chicken sits near the top of that range at roughly 31 g of protein per 100 g of food. Egg white sits near the bottom at roughly 11 g per 100 g and is about 88 percent water; Greek yogurt is close to 10 g per 100 g, tofu close to 8 g, cooked chickpeas close to 9 g. To reach the same 30 g floor, a slot built from those ingredients has to weigh three to six times as much, and nearly all of the additional weight is water. Grouped by ingredient, the pattern is consistent rather than anecdotal. The 52 recipes containing chicken average a 118 g protein slot, 34.5 g of protein, and a yield of 0.31. The 98 recipes containing egg white average a 265 g slot, 32.8 g of protein, and a yield of 0.13. The heavier group carries 147 g more food and delivers 1.7 g less protein.
What Failed
Three things, and the first is the largest. Yield is not a lever a formulator gets to pull. All 52 chicken-containing recipes sit in the WAI, WAY, or WI body types; all 98 egg-white recipes sit in EI, EY, MI, or MY. The two groups do not overlap. That means the six-fold yield difference is very nearly a restatement of which body type the recipe was written for, not evidence that one formulation approach is more efficient than another. We cannot claim that high-yield recipes are better formulated, because no recipe in the database was ever free to choose. The protein family is fixed upstream by the body type, and yield follows from it. Reading this as a formulation quality signal would be reading a confound. Second, our compensation story is weaker than we wanted it to be. If the extra 147 grams in the egg-white group were buying fiber, we would expect a meaningful fiber gap. The gap is 7.72 g against 6.88 g — 0.84 g, against a database average of 7.62 g. The correlation between slot mass and fiber is only 0.21. Fiber does not explain 147 grams of food. Water does, and water has no macro line of its own. Third, our internal Score does not track yield in any way that would justify treating yield as a quality measure. The correlation is 0.29. If protein efficiency mattered as much as the framing of this note implies, the Score should reward it clearly, and it does not. We are reporting that rather than quietly dropping it.
What We Learned
Slot weight is a proxy for water and food matrix, not for protein. The number a formulator has been reading as "how much protein is in here" has been reading, for most of the database, as "how wet is this ingredient." Once yield is computed explicitly, the 60-gram slot and the 365-gram slot stop looking like a mistake and start looking like two different ingredients doing the same job through different amounts of water. The second lesson is that the heavy slot is not waste. Holt's data is direct on this point: at matched calories, serving weight and water content were the two strongest positive correlates of satiety, at r = 0.66 and r = 0.64. Our two endpoint shakes are 2.8 kcal apart and differ by 305 grams of food. By that evidence the heavier one is doing something the lighter one is not — it is just not doing it in the protein column. And Berrazaga's review gives an independent reason not to trim it: plant-source protein carries a lower anabolic yield per gram, and a higher dose plus blending of multiple plant sources are the documented ways to compensate. Chickpeas, soy milk, black beans and egg white together are that strategy written out as a recipe.
What Changed at MOMS
Formulation sheets now carry a computed yield column — grams of protein per gram of slot — alongside slot mass, so the 30 g protein floor is checked against composition rather than against the size of the scoop. We stopped treating a heavy protein slot as something to trim on sight, because trimming it removes water and volume, which is the part Holt's data says is carrying satiety at matched calories. We replaced the old heuristic with a narrower flag. A protein slot is now reviewed only when yield falls below 0.09 and fiber falls below the 7.62 g database average at the same time — the one case where the extra mass is buying neither protein nor fiber and is therefore genuinely idle. Everything above that line is left alone. And because yield is set by body type rather than chosen, we removed it from our internal recipe-quality comparisons entirely; comparing a WAI recipe's yield to an EY recipe's yield was comparing two different questions.
References
Berrazaga I, Micard V, Gueugneau M, Walrand S. The Role of the Anabolic Properties of Plant- versus Animal-Based Protein Sources in Supporting Muscle Mass Maintenance: A Critical Review. Nutrients. 2019;11(8):1825. | Holt SHA, Miller JCB, Petocz P, Farmakalidis E. A satiety index of common foods. European Journal of Clinical Nutrition. 1995;49(9):675-690.
Keep reading
Previous experiment
Not sure which shake fits you? We match the formulation to your body type.
Find your body type