In this LAB note
Question · Hypothesis · Method · Results · What Failed · What We Learned · What Changed at MOMS
Question
Why does the Metal body type get 37% fewer shake recipes than Earth?
Hypothesis
We expected the gap to be a scheduling artifact — Metal-type users seeing fewer weeks in the rotation, or fewer Calorie-Label and Goal categories to choose from, the same kind of quiz-side thinning we've documented before (LAB #026). We did not expect an ingredient-level explanation.
Method
We pulled all 201 recipes from the internal recipe database and grouped them by the eight body-type (BT) codes: Earth-Yang (EY), Earth-Yin (EI), Wood-Yang (WY), Wood-Yin (WI), Metal-Yang (MY), Metal-Yin (MI), Water-Yang (WAY), and Water-Yin (WAI). For each code we counted total recipes, the distinct weeks represented, the distinct Calorie-Label and Goal categories offered, and — by splitting each recipe's Liquid and Protein fields on their "+" delimiter — the distinct base ingredient names actually used. We cross-checked the finding against dietary-diversity research: McDonald et al.'s American Gut Project (2018) found that people eating more than 30 distinct plant types a week carry measurably more diverse, short-chain-fatty-acid-producing gut microbiota than people eating fewer than 10 — evidence that the number of distinct plant-derived inputs in a diet is not a cosmetic detail.
Results
MY and MI each carry 19 recipes; EY and EI each carry 30 — Metal's menu is 37% smaller, code for code. It isn't the rotation: all eight codes span the same four weeks, and every code offers the same 3-4 Calorie-Label categories and the same 4 Goal categories. It's the ingredient list. EY, EI, WY, WI, and WAY all draw their liquid base from the same five ingredients: Almond Milk, Coconut Milk, Coconut Water, Oat Milk, and Soy Milk. WAI drops one (Coconut Water). MY and MI drop three at once — Oat Milk, Soy Milk, and Almond Milk appear zero times across their combined 38 recipes. In their place: Rice Milk, which appears in 28 recipes sitewide, and every single one of them is Metal (e.g. MY-W1-COOL-010-V1 "Cool Ease," MI-W1-BUILD-003-V1 "Foundation Fuel") — versus EY-W1-ENERGY-006-V1 "Vital Spark" (Soy Milk 300ml) or EY-W1-ENERGY-011-V1 "Amber Rise" (Almond Milk 220ml), recipes a Metal-type quiz result never returns. The protein slot shows the same pattern at smaller scale: combined across both Metal codes, only 7 distinct protein sources appear (Adzuki Beans, Chickpeas, Egg, Egg White, Kidney Beans, Lentils, Tofu); combined across both Earth codes, 12 do, adding Black Beans, Cashew, Greek Yogurt, Soy Milk, and White Beans to the list.
What Failed
Our scheduling hypothesis failed outright — weeks, Calorie-Label spread, and Goal spread are identical across all eight codes, so none of them explain the gap. A second expectation also failed: we assumed a diet built on Rice Milk and Coconut Milk instead of Soy Milk and Oat Milk would show up as lower protein, since soy milk carries meaningfully more protein per serving than rice milk. It didn't. Average protein lands at 33.4g (MY) and 32.9g (MI) versus 32.2g (EY) and 33.0g (EI) — Metal is not behind on protein; if anything it's marginally ahead, because the protein slot (egg white, tofu, lentils) was pushed harder to compensate. The Score system also failed to register any of this: average Score is 14.8 for Metal and 14.8 for Earth — a formulation constraint this large is invisible to the metric we use to rank recipes. What the swap does cost is fiber: 6.35g average for Metal versus 8.6g for Earth, a real 26% reduction that traces to the liquid slot rather than the produce slot.
What We Learned
A body type's menu size isn't a quiz-logic artifact — it's the direct, countable output of how many raw ingredients that type's rules permit. Metal loses three liquid bases and roughly five protein sources compared to Earth, and across a four-week, multi-Goal, multi-Calorie-Label grid, that compounds into a menu that's a third smaller by the time a customer sees it. The Score system, built to rank recipes on macros, has no mechanism to see or correct for this kind of structural narrowing — it would need a separate check.
What Changed at MOMS
We flagged Metal-Yang's Week 4 (2 recipes, the thinnest slot in the entire database) for priority recipe development, and we're defaulting a fiber-forward functional add-in — chia or flaxseed — into new Metal recipes specifically, since the fiber shortfall traces to the excluded liquid bases, not the produce slot, and won't fix itself through produce substitution alone.
References
McDonald D, Hyde E, Debelius JW, et al. American Gut: an Open Platform for Citizen Science Microbiome Research. mSystems. 2018;3(3):e00031-18.
Keep reading
Previous experiment
Not sure which shake fits you? We match the formulation to your body type.
Find your body type