In this LAB note
Question · Hypothesis · Method · Results · What Failed · What We Learned · What Changed at MOMS
Question
Does a longer ingredient list put more protein in the blender? Two of our recipes sit 0.8 calories apart: one runs on thirteen ingredients, the other on five. We wanted to know whether the count itself tracks protein across all 201 formulations, or whether it tracks something else entirely.
Hypothesis
We expected a weak positive relationship. Every additional ingredient is another opportunity to add protein, and our longest recipes stack the protein slot with three or four sources instead of one. We predicted that recipes at the top of the ingredient-count range would average two to three grams more protein than recipes at the bottom, a correlation somewhere near r = 0.2: small, but pointing the right way.
Method
Published research first. The food-variety literature is consistent that adding items to a meal changes how much a person eats, not how much nutrition the meal delivers. Rolls and colleagues (1981) served 36 subjects sandwiches with either one filling or four; the four-filling group ate about a third more, and a parallel study using three distinct yogurt flavors produced the same effect even when the single-flavor condition offered the subject's own favorite. A 2021 systematic review and meta-analysis by Embling and colleagues screened 7,259 references, reviewed 34 articles, and pooled 30 studies (39 comparisons) in a random-effects model. Greater variety was associated with greater consumption, Hedges' g = 0.405 (95% CI: 0.259, 0.552). Variety moves intake. Neither paper claims it moves nutrient density. Then the internal analysis. We counted discrete ingredients in each of our 201 production recipes across all six formulation slots (Liquid, Body, Fruit & Vegetable, Protein, Functional, Flavor), treating each named item carrying its own gram weight as one ingredient. We then correlated that count against Kcal, Protein, Carbohydrate, Fat, Fiber, and our internal Score. Finally we pulled the two recipes at opposite ends of the count range whose calorie totals sat closest together, so the head-to-head comparison would hold energy nearly constant.
Results
Ingredient counts run from 5 to 13, mean 8.5. Ingredient count versus protein: r = -0.08. That is not a weak positive relationship. That is no relationship, tipped very slightly the wrong way. The extreme pair, less than one calorie apart. WAI-W1-WARM-014-V1 ("Sunwarmed Flow"), 5 ingredients: Almond Milk 290 ml, Avocado 55 g, Spinach 30 g, Dates 32.5 g, Chicken 105 g. It lands at 395.9 kcal, 36.3 g protein, 31.3 g carbohydrate, 15.4 g fat, 7.6 g fiber. EY-W3-GUT-004-V1 ("Belly Balance"), 13 ingredients: Oat Milk 97.5 ml, Soy Milk 97.5 ml, Beet 25 g, Raspberry 35 g, Dates 13 g, Egg 27.5 g, Chickpeas 30 g, Adzuki Beans 15 g, Egg White 150 g, Chia Seeds 5 g, Cashew 6.3 g, Monk Fruit 1 g, Agave 5 g. It lands at 396.7 kcal, 30.8 g protein, 44.2 g carbohydrate, 11.6 g fat, 10.3 g fiber. Eight more ingredients. 0.8 more calories. 5.5 grams less protein. The group means run the same direction. Six-ingredient recipes (n = 20) average 32.8 g protein; twelve-ingredient recipes (n = 5) average 32.2 g; thirteen-ingredient recipes (n = 3) average 31.6 g. Doubling the list costs about a gram. The mechanism is a ceiling, not a preference. Every recipe is built to land under 400 kcal while clearing 30 g of protein. Once the protein floor is covered, every additional ingredient has to be paid for out of the same calorie budget, and the things a formulator reaches for next (texture, color, gut function, sweetness) are carbohydrate and fat, not protein. EY-W3-GUT-004-V1 spends four separate items and 222.5 g of food on its protein slot and still finishes below one 105 g piece of chicken, because those four items drag 44.2 g of carbohydrate in with them. The variable that does track ingredient count is fat: r = +0.37, the strongest correlation in the set. Six-ingredient recipes average 7.4 g of fat; twelve-ingredient recipes average 14.6 g. A longer list buys fat.
What Failed
Four things, and the first is the headline we did not get to write. "Shorter is better" is not what this data says. Fiber moves in the opposite direction from protein: 7.1 g at six ingredients, 10.0 g at twelve, r = +0.16. EY-W3-GUT-004-V1 carries 10.3 g of fiber against WAI-W1-WARM-014-V1's 7.6 g. The thirteen-ingredient recipe is not a worse shake. It is a differently optimized one, and it wins on the axis it was built for. Second, our own Score does not support the argument. Ingredient count versus Score is r = +0.09, which is nothing. If ingredient economy were a virtue our scoring system recognized, it would show up here, and it does not. It is worse than neutral: the thirteen-ingredient Belly Balance scores 16, while WI-W3-HEAL-008-V1 ("Warm Comfort") - 6 ingredients, 35.4 g protein, 13.8 g fiber, 395.4 kcal - scores 12, near the bottom of our 11-to-20 range. Whatever the Score is measuring, it is not this, and we are not going to imply that it is. Third, the tails are thin. Only three recipes carry 13 ingredients and only one carries 5. The r = -0.08 across all 201 is the durable result; the six-versus-thirteen group means rest on 20 recipes and 3. We are reporting a direction, not a dose-response curve, and one more thirteen-ingredient recipe could move that group mean by more than a gram. Fourth, we could not test the thing the published research actually measures. Rolls and Embling study how much people eat when offered variety. A MOMS shake is a fixed 16 oz serving, so that mechanism has nowhere to act. We cite those papers as a reason to be skeptical of ingredient count as a quality signal, not as evidence about what our own customers drink.
What We Learned
Ingredient count is a record of how the leftover calories were spent, not a measure of how much protein arrived. In a formulation with a hard 400-kcal ceiling and a hard 30 g protein floor, protein is the constrained variable and everything else is the free one. That is how a five-item recipe and a thirteen-item recipe end up 0.8 calories apart with 5.5 grams of protein between them, in the direction nobody predicts. The general version: a long ingredient list is a sensory and functional decision. Reading it as a nutrition decision is the error. The published variety work arrives at the same place from the other side, finding that variety reliably moves how much people eat (g = 0.405) while saying nothing at all about what is in what they are eating.
What Changed at MOMS
We stopped treating ingredient count as a quality signal in recipe review, in either direction. It had never been written down as a rule, which is exactly why it needed removing: reviewers were reading a four-source protein slot as more thorough than a single-source one. Protein-slot review is now done on grams of protein returned per gram of protein-slot food, evaluated as one line rather than item by item. Under that check, EY-W3-GUT-004-V1's four items read as 222.5 g of food returning 30.8 g of protein, or 0.14 g per gram, against WAI-W1-WARM-014-V1's 105 g returning 36.3 g, or 0.35 g per gram. Both recipes stay on the menu. They are now compared on the same number. We also added one rule for the Functional and Flavor slots: an ingredient entering below 7 g has to justify itself on a named function, not on appearing in the list. That is where long recipes accumulate, and it is the one place where the calorie cost is smallest and the protein return is exactly zero.
References
Embling R, Pink AE, Gatzemeier J, Price M, Lee MD, Wilkinson LL. Effect of food variety on intake of a meal: a systematic review and meta-analysis. The American Journal of Clinical Nutrition, 113(3), 716-741, 2021. | Rolls BJ, Rowe EA, Rolls ET, Kingston B, Megson A, Gunary R. Variety in a meal enhances food intake in man. Physiology & Behavior, 26(2), 215-221, 1981.
Keep reading
Previous experiment
Not sure which shake fits you? We match the formulation to your body type.
Find your body type