#009
In this LAB note
Question · Hypothesis · Method · Results · What Failed · What We Learned · What Changed at MOMS
Question
In a whole-food shake with a fixed calorie budget and a protein target already met, which macronutrient actually predicts a better formulation: fat or fiber?
Hypothesis
We expected fiber to be the stronger signal. Fiber is the macronutrient we have historically protected first: it slows digestion, feeds the gut, and is the easiest thing to point at when explaining why a whole-food shake beats a powder. Fat, by contrast, we treated as a cost - calorie-dense, easy to overshoot, and useful mostly for texture. So we predicted a positive correlation between fiber and our internal Score, and something close to zero, or slightly negative, for fat.
Method
Two published lines framed the test. Kdekian and colleagues (European Journal of Clinical Nutrition, 2020) meta-analyzed ten randomized trials in which fat and carbohydrate were exchanged isocalorically in mixed meals, holding protein comparable - the same constraint our recipes operate under. Each 10% of energy shifted from carbohydrate to fat lowered mean postprandial glucose by 0.32 mmol/l and postprandial insulin by 18.2 pmol/l. Jenkins and colleagues (British Medical Journal, 1978) supplied the other half: across guar, pectin, tragacanth, methylcellulose, wheat bran, and cholestyramine at a matched 12g of fibre, the reduction in peak blood glucose tracked each substance’s viscosity (r = 0.926), not its mass. Fiber quantity, in other words, is not the active variable. Internally, we took all 201 recipes in our formulation database - mean 375 kcal, mean 33.0g protein - and correlated Score against each macro, then binned recipes by fat and by fiber and compared mean Score within each bin. We also matched recipes on Goal, calories (within 8 kcal), and protein (within 2.5g) to isolate the fat-for-carbohydrate swap.
Results
Fat was the strongest macro predictor of Score in the database (r = +0.209), ahead of protein (r = +0.070) and calories (r = +0.022). Fiber ran the other way (r = -0.171). Binned, the pattern is monotonic and then flat: recipes under 7g of fat averaged a Score of 14.31 (n=39), 7-10g averaged 14.76 (n=62), 10-13g averaged 15.45 (n=53), and 13-16g held at 15.41 (n=32). Fiber inverted it - under 7g of fiber averaged 15.37, while 9-11g averaged 14.38 (n=48). The mechanism is displacement. Fat and carbohydrate are strongly inversely correlated across the database (r = -0.708) because the calorie ceiling is effectively fixed; fiber and carbohydrate move together (r = +0.522) because in a whole-food shake most fiber arrives inside fruit that also carries sugar. Fiber is not hurting the recipe. It is riding along with the thing that is. Two matched pairs make it concrete. Same Goal, near-identical calories and protein: WI-W3-HEAL-001-V1 (Soothing Sip) runs 394.2 kcal, 33.4g protein, 12.6g fat from Avocado 39g and Chia Seeds 10g, 39.1g carbohydrate, 12.1g fiber - Score 17. Its sibling WI-W3-HEAL-008-V1 (Warm Comfort) runs 395.4 kcal, 35.4g protein, but only 4.0g fat, with carbohydrate at 56.9g from Beet 84.5g, Blueberry 59.1g, and Dates 10.2g - Score 12. It carries more fiber (13.8g) and scores five points lower. The same shape repeats in EY-W2-DETOX-002-V1 (Soft Reset, 14.6g fat, 32.3g carb, Score 17) against EY-W2-DETOX-008-V1 (Weightless Reset, 6.8g fat, 48.5g carb, Score 13) - identical Coconut Milk 300ml base, 8 kcal apart.
What Failed
Three things. First, the ceiling. Above 16g of fat the gain reverses: those 15 recipes average 15.07, below the 13-16g band. EY-W4-CALM-007-V1 (Soft Landing) carries 17.9g of fat on Egg 110g and Kidney Beans 46.5g and scores 13; WY-W1-DETOX-013-V1 (Clarity Flow) carries 17.7g and also scores 13. Past roughly 16g, fat stops displacing carbohydrate and starts displacing the volume that fruit, vegetables, and functional ingredients need. Second, the naive fix. If fat is good, adding a fat ingredient should be good - so we split the database by whether a recipe contains avocado, nuts, seeds, or nut butter. The 125 recipes that do average a Score of 15.02; the 76 that do not average 14.91. Effectively nothing. It is not the ingredient. It is what the ingredient displaces. Third, our own follow-up idea. We hypothesized that fiber density - fiber per gram of carbohydrate - would rescue fiber as a predictor and separate fiber from kale from fiber from dates. Correlation with Score: -0.007. Flat. The ratio told us nothing the raw carbohydrate number had not already said.
What We Learned
Fat is not a cost line in a fixed-calorie shake; it is the cheapest way to buy carbohydrate space. And fiber, on its own, is a poor proxy for formulation quality - not because fiber is unhelpful, but because in whole-food recipes the fiber number is largely a shadow of the fruit load. Jenkins showed in 1978 that viscosity, not grams, does the work; our database says the same thing from the opposite direction. Counting fiber grams measures how much fruit is in the blender, which is not the question we thought we were asking.
What Changed at MOMS
We stopped treating fiber as a target and started treating it as a readout. Recipes are now built to a fat band of 10-16g, with the lower bound enforced on any recipe carrying more than 45g of carbohydrate; below 7g of fat a recipe gets flagged for review rather than passed on protein alone. Fat above 16g now requires a justification tied to the Goal, not to texture. And we no longer credit a recipe for high fiber when that fiber arrives with dates, dried fruit, or more than two high-sugar fruits - in those cases we record the carbohydrate first and ask what a swap into avocado, chia, or a tree nut would buy.
References
1. Kdekian A, Alssema M, van der Beek EM, Greyling A, Vermeer MA, Mela DJ, Trautwein EA. Impact of isocaloric exchanges of carbohydrate for fat on postprandial glucose, insulin, triglycerides, and free fatty acid responses - a systematic review and meta-analysis. European Journal of Clinical Nutrition. 2020;74(1):1-8. 2. Jenkins DJA, Wolever TMS, Leeds AR, Gassull MA, Haisman P, Dilawari J, Goff DV, Metz GL, Alberti KGMM. Dietary fibres, fibre analogues, and glucose tolerance: importance of viscosity. British Medical Journal. 1978;1(6124):1392-1394.
Keep reading
Previous experiment
Next experiment
Not sure which shake fits you? We match the formulation to your body type.
Find your body type