How the Gut Score works, rule by rule.
Most food scores are black boxes. Ours is not. Here is every rule the analysis applies, why it exists and where the evidence comes from.
- 40
- where every meal starts
- 0–100
- the range, used as is
- 4
- things that move it
01
The meal score
Every meal starts at 40 and moves for four reasons. The total is clamped between 0 and 100 and shown exactly as calculated, so a gut-hostile meal really does read low.
A neutral meal sits below the middle on purpose. A plate has to earn its way up.
Fiber in grams from the whole meal.
Live probiotic food +14, other fermented food +8, legumes +10, whole grains +8, nuts and seeds +5, polyphenol-rich foods +5, prebiotic foods +5, herbs and spices +3.
Each whole or minimally processed plant in the meal.
Ultra-processed food or processed meat −25, deep-fried −20, added sugar −15, refined grains −10, weighted by how much of the meal they make up.
Rating words
- Excellent 85–100
- Great 65–84
- Average 50–64
- Okay 30–49
- Low fuel 0–29
The lowest word is “Low fuel”, never “bad”. Food is not moral, it is just more or less useful to your microbes. Large portions never inflate the score. A bigger plate of the same food does not get a better grade.
02
Why these rules
- Fiber carries the most weight
- Intakes of 25 to 29 grams a day or more were linked with the largest reductions in heart disease, stroke, type 2 diabetes, colorectal cancer and mortality in a WHO-commissioned Lancet review.1
- Live ferments beat pasteurized ones
- Fermented is not the same as live. A Stanford trial found that a diet rich in live fermented foods increased microbiome diversity and lowered inflammatory markers, so live cultures earn +14 and pasteurized, baked or cooked ferments +8.3,4
- Variety counts separately
- Plant variety was linked with microbiome diversity in the American Gut Project, and different fibers and polyphenols feed different microbes. That is why gut fuel classes and whole plants each add points.2,7,6
- Friction is weighed, not banned
- High ultra-processed food intake is linked with a wide range of poorer outcomes, and the WHO recommends limiting free sugars. Friction is scaled by how much of the meal it makes up, so a little ketchup is not a currywurst.5,8
- Processing matters, even for plants
- In the PREDICT 1 study, minimally processed plant foods were associated with favorable gut microbes and more processed plant foods were not. Fries do not earn plant points.9
03
The daily score
Each analysis sees one meal, so the app computes your day itself. The daily score fills toward your gut goals for the day.
Hit every goal and the day reads 100.
Good food is retained and additive. Fiber you ate in the morning is still fermented into short-chain fatty acids hours later, so one heavy meal never erases it. Friction is a bounded drag: the friction classes present that day are added up, capped at 30 points and scaled by friction food’s share of the day’s grams. A single sausage or soda among good meals costs a few points. A day of only fast food drags fully.
Example: a great breakfast followed by currywurst and fries lands in the mid-40s, not the teens.
04
Plant points
Plant points feed your weekly 30 and your daily variety. A food earns one only when it is a whole or minimally processed plant: the analysis identifies it as a plant and it carries no friction class. A whole rye slice counts, a bagel made from refined flour does not, and neither do fries. Mixed products like muesli count by their ingredients, never as one plant. Each plant counts once per week, and herbs and spices count too.
Read: 30 plants a week, explained05
The inflammation read-out
Every meal also gets a separate score from −5 to +5, starting at 0. Omega-3 fats are the headline, with EPA and DHA from oily fish or algae strongest and ALA from walnuts, flax and chia milder. Extra virgin olive oil, turmeric, ginger and polyphenol-rich foods move it up. Processed and red meat in larger amounts, deep frying and a heavy sugar load move it down. It has its own gauge and never changes the gut score, so the core score stays anchored to fiber, variety and ferments.
06
How the analysis works
-
Photo, barcode or text
A language model identifies each food, its portion in grams, its fiber in grams and its classes, then applies the rules above with judgment.
-
Barcodes use real label data
For packaged food, fiber comes from the product’s Open Food Facts entry, and the model scores the rest.
-
A deterministic fallback
If the analysis is unavailable, a fixed implementation of the same rules scores the meal, so results stay consistent.
-
Photos are not kept
A photo goes to our server for the analysis and is deleted right after.
07
Limits
- Portion and fiber values are estimates from a photo or a description, not laboratory measurements.
- The score reflects research on populations. Your own gut may respond differently, which is what the daily check-in is for.
- Feed Your Gut is not a medical device and does not diagnose or treat anything. If you have IBS, IBD, coeliac disease or another condition, follow your clinician’s advice first.
- Protein and calories are shown as neutral estimates and are never part of the gut score.
Questions about the score
Why does every meal start at 40 and not 50?
A neutral plate should not look good by default. Starting at 40 means a meal has to bring fiber, gut fuel or whole plants to reach the upper range, which makes the score more informative.
Why is fiber capped at 30 points?
Without a cap, one bowl of bran could max out the score on fiber alone. The cap keeps room for variety and ferments to matter, because a gut-friendly meal is more than one nutrient.
Does a bigger portion get a better score?
No. Large portions never inflate the score. The score describes the quality of what is on the plate, not the amount.
Why can a real scan differ from these rules by a few points?
The analysis applies the rules with judgment, for example when deciding how much of a mixed dish is fried or refined. Small deviations are expected; the rules are the anchor.
Sources
- Reynolds et al., The Lancet 2019
- McDonald et al., mSystems 2018
- Wastyk et al., Cell 2021
- Marco et al., Nat Rev Gastroenterol Hepatol 2021
- Lane et al., BMJ 2024
- Cardona et al., J Nutr Biochem 2013
- Holscher, Gut Microbes 2017
- WHO, Guideline: sugars intake for adults and children, 2015
- Asnicar et al., Nature Medicine 2021