Turning nutrition into a grade you can trust
Cado gives every recipe a letter grade per serving, plus goal lenses like Heart Health, Blood Sugar, and Less Processed. Each grade is built from separate factors, each judged on its own criteria. The grade was the easy part. The work was making it honest, hard to game, and able to explain itself.
- Problem
- Nutrition data is dense, and most food scores are easy to game: stack enough protein and the sugar disappears.
- Build
- A layered engine that measures each recipe, judges it against published guidelines, and explains the result down to the ingredient.
- Principle
- Measure once, judge per goal. Scales follow published guidelines wherever they exist, and when the data isn't good enough, there's no grade.
Inputs
Composition
Per servingNutrientsIngredientsGuidelines
WHODASHData quality
Coverage check
Cado score
Salmon grain bowl
- Free sugarGreat
- NutrientsGood
- SodiumWatchModerate
- Data coverage Enough to grade
Action
Suggested swap
Soy sauce → low-sodium soy sauce
Sodium: Moderate → Good
Ranks in
Your Heart Health feed
In the app
01Grade + reasons
Next to the grade, the reasons behind it, plus swaps that call out the ingredient driving each factor.
02Factor breakdown
Each factor sits on its own scale, citing the guideline it comes from (WHO, DASH, FDA) where one exists.
03How it's measured
Every factor explains itself in plain language, cutoffs included.
04Goal lenses
Same facts, different lens: pick the goal that matters to you.
Measure, then judge
- Ingredients are resolved to a food database, and nutrition is computed per serving
- Measurement happens once. Each score then applies its own judgment to the same facts
- Data comes from multiple sources, each tracked with where it came from and how confident it is
Hard to game
- The overall score balances nutrient density against saturated fat, added sugar, sodium, and processing
- Factors are combined so that one weak spot can't be hidden by strengths elsewhere
- Sugar is judged by where it comes from, so fruit isn't treated like syrup
Grounded in guidelines
- Where a published guideline exists (WHO, DASH, FDA), a factor's scale is anchored on it and cited in the app
- Hard caps only apply where a guideline supports them, and they should fire rarely
- Grade cutoffs were calibrated so the labels match what the guidelines would say
Refuse rather than guess
- If too little of a recipe is backed by real nutrition data, it shows the facts but no grade
- A missing score never ranks
- Every grade comes with its reasons, never just a letter
Explaining the why
- Each factor shows where the recipe lands and the guideline behind it
- The ingredients driving a weak factor are named, along with what changes without them
- One tap hands those ingredients to the assistant to swap or reduce
Goals without personalizing the score
- The score is the same for everyone, so it stays comparable and checkable
- Personalization is choosing the lens: your health goals decide what the feed ranks first
- Copy is written about nutrients, never framed around a medical condition
Validation
- Grades are checked to spread across every level instead of piling up in the middle
- Score rankings are compared against dietitian-style judgments and hand-sorted examples
- Next: clinical review of the medically named lenses
