In this LAB note
Question · Hypothesis · Method · Results · What Failed · What We Learned · What Changed at MOMS
Question
Does a body type with more recipes in its rotation actually get more different plants?
Hypothesis
We expected plant variety to scale with catalog depth. Each of our eight body types draws from its own subset of the 201 recipes, and those subsets are not the same size: some body types have 30 recipes to rotate through, others have 19. Every additional recipe is another chance to introduce a plant the others do not use, so a 30-recipe rotation should end up meaningfully wider than a 19-recipe one. If that held, the fix for uneven variety would be simple and purely arithmetic. Write more recipes for the smaller body types and the gap closes on its own.
Method
Published research first. McDonald and colleagues, working with more than 10,000 stool samples in the American Gut Project, found that the number of distinct plant types a person eats predicts gut microbial diversity better than categorical diet labels such as vegan or omnivore. Comparing participants eating fewer than 10 plant types per week against those eating more than 30, they measured a significant increase in 16S alpha-diversity in the high-diversity group (Kruskal-Wallis H = 65.817). Heiman and Greenway make the mechanistic argument: each plant substrate gives a competitive advantage to a different set of gut species, so a narrower substrate list supports a narrower and less adaptable community. Then our own catalog. We pulled all 201 recipes and extracted every ingredient from the six formulation slots: Liquid, Body, Fruit & Vegetable, Protein, Functional, Flavor. That yields 74 distinct ingredient names. Ingredient names are not species, so we normalized. Almond milk and almonds are one plant. Oat milk and oatmeal are one plant. Soy milk, tofu and silken tofu are one plant. Coconut milk and coconut water are one plant. Rice milk and white rice are one plant. The same collapse applies to apple and green apple, cacao and cocoa, lentils and green lentils, and ginger and fresh ginger. We then removed the five animal-derived ingredients in the catalog: egg, egg white, chicken, Greek yogurt and honey. Seventy-four ingredient names collapse to 59 distinct plant species. For each body type we counted how many of those 59 appear anywhere in its recipes, then divided that body type total plant slot-fills by its unique species count to get a repetition rate: how many times the average plant reappears across the rotation.
Results
Recipe count barely predicts species count. Across the eight body types the correlation is 0.53 — real, but nothing like the near-lockstep relationship we assumed. The cleanest pair is WY against WAY. WY draws on 28 recipes and 203 plant slot-fills and reaches 39 distinct species. WAY draws on 26 recipes and 189 slot-fills — two fewer recipes and fourteen fewer ingredient placements — and reaches 47 species. Eight more plants from the smaller menu. EY and EI are the two largest rotations in the catalog at 30 recipes each, and both land on 36 species, eleven short of WAY. Repetition is the variable doing the work. Plant slot-fills divided by unique species, by body type: EY 5.78, EI 5.75, WI 5.33, WY 5.21, MY 4.59, WAY 4.02, WAI 4.00, MI 3.94. The two body types with the most recipes have the most repetitive ones. You can see it inside a single glass. MI-W3-VITAL-008-V1 (Bright Vitality, Standard, 387 kcal, 33.4 g protein, 5.8 g fiber) is Rice Milk 140 ml, White Rice 55 g, Papaya 70 g, Lentils 55 g, Egg White 240 g, Agave 10 g. Six ingredient lines, but rice milk and white rice are the same plant, so it delivers 4 plant species. WAY-W1-HYDRATE-002-V1 (Deep Quench, Filling, 395.5 kcal, 33 g protein, 8.4 g fiber) is Oat Milk 150 ml, Avocado 17.5 g, Spinach 20 g, Green Apple 30 g, Tofu 50 g, Walnut 5 g, Chickpeas 30 g, Chicken 60 g, Flaxseed 4 g, Maple Syrup 5 g: nine plant species, within nine calories and half a gram of protein of the other shake. The pattern scales up to the cell. MI Week 3 holds four recipes that between them fill 21 plant slots with only 10 distinct species: rice, chickpeas, lentil, chia, turmeric, agave, papaya, kale, cabbage and dates. WAY Week 1 holds ten recipes and reaches 36 species. Every recipe in both cells clears our 30 g protein floor.
What Failed
Our Score does not track any of this. Mean Score by body type against species count gives a correlation of 0.42, and the ordering breaks exactly where it matters. WY has 39 species, the third-widest rotation we build, and the lowest mean Score of any body type at 13.8. MY has 29 species, the narrowest rotation we build, and scores higher at 14.6. If plant-species breadth were something the Score measured, that pair could not exist. We should also be plain about what this analysis does not show. The American Gut result is an observational association in a self-selected cohort, not evidence that adding a plant to a shake changes anyone microbiome, and we have measured nothing about our own customers gut health. Nothing here demonstrates that a WAY rotation produces a better outcome than an MY rotation. What we have shown is narrower and entirely internal: our system produces an 18-species spread across body types that nobody designed and nothing in our tooling was watching. And recipe count is not irrelevant, which was the half of the hypothesis that survived. The two 19-recipe body types, MY and MI, sit at 29 and 31 species, the bottom of the range. Catalog depth sets a ceiling. Repetition decides how close you get to it.
What We Learned
Variety in a personalized system is two quantities, and we had been tracking only one. How many recipes a body type is offered sets an upper bound on plant exposure. How often those recipes reuse the same plant determines where inside that bound you actually land, and in our current catalog it is the larger of the two effects. A body type can gain recipes and lose ground.
What Changed at MOMS
We added a rotation-level diversity check that runs on the body-type subset rather than on the individual recipe. Two numbers now sit beside the Score whenever a recipe is added or edited: distinct plant species across that body type full rotation, and that rotation repetition rate. Species are counted after the same normalization used here, so oat milk and oatmeal cannot both earn credit. Any new recipe written for a body type at or above a 5.0 repetition rate has to introduce at least one species that rotation does not already contain. The near-term target is every body type at 36 species or more with no rotation above 5.0, which means the next recipes we write for EY, EI, WI and WY are constrained to unused plants rather than to new arrangements of the plants already in rotation.
References
McDonald D, Hyde E, Debelius JW, et al. American Gut: an Open Platform for Citizen Science Microbiome Research. mSystems, 3(3):e00031-18, 2018. | Heiman ML, Greenway FL. A healthy gastrointestinal microbiome is dependent on dietary diversity. Molecular Metabolism, 5(5):317-320, 2016.
Keep reading
Not sure which shake fits you? We match the formulation to your body type.
Find your body type