A tablespoon of butter and a tablespoon of olive oil look nearly identical on a nutrition label: roughly the same calories, roughly the same total fat. The label's total-fat line treats them as equals. Your cardiovascular system does not. Butter's fat is about two-thirds saturated; olive oil's is overwhelmingly unsaturated. And decades of trials and cohort studies point the same direction: replace saturated fat with unsaturated fat and LDL cholesterol falls, along with cardiovascular risk.
This is the blind spot the low-fat era never fixed. For years the advice was simply "eat less fat," and food companies obliged by stripping fat out and backfilling with refined starch and sugar. The evidence has since moved on. How much fat you eat matters less than which fat it is — and no single line on the label answers that question.
So Nutrication grades fat by composition, not just quantity. Here is how the three fat families differ, how one ratio captures most of the story, and why the app sometimes refuses to rate fat quality at all.
The three fat families
Saturated fat is the kind that stays solid at room temperature — concentrated in butter, cheese, cream, fatty red meat, and coconut and palm oil. It reliably raises LDL cholesterol, which is why major health bodies, including the World Health Organization, recommend limiting it and replacing it with unsaturated fat. Limiting, not banning: dose and context matter, and a slice of cheese is not a stick of butter.
Unsaturated fat comes in two branches. Monounsaturated fat dominates olive oil, avocados, and most nuts. Polyunsaturated fat includes the omega-6 fats of seed oils and the omega-3s of fatty fish, walnuts, chia, and flax. In trial after trial, swapping saturated fat for either branch improves blood lipids. That substitution effect is one of the least contested findings in nutrition science.
Trans fat is the outlier — bad enough to earn its own section.
Trans fat: the one hard no
Industrial trans fat is created by partially hydrogenating liquid oils to make them solid and shelf-stable. It raises LDL and lowers HDL — the worst of both directions, and the reason there is no defensible intake level. The FDA revoked its "generally recognized as safe" status in 2015, with a primary compliance date of June 2018, and the WHO has campaigned to eliminate industrially produced trans fat from the global food supply.
Two label subtleties are worth knowing. First, "fully hydrogenated" oil is not the same red flag — full hydrogenation produces essentially no trans fat. Second, the trans-fat line itself can read zero while the product still carries a little: US rules allow amounts under half a gram per serving to round down, and emulsifiers like mono- and diglycerides are exempt from that line entirely. The ingredient list is the more honest witness, and the phrase to scan for is partially hydrogenated. There is more on decoding labels in how to read a nutrition label.
One number that captures it: the unsaturated share
Nearly every packaged food reports total fat and saturated fat — those two fields exist on roughly 96% of products in the database Nutrication draws from. Subtract saturated and trans fat from the total, and what remains is unsaturated. Divide by total fat and you get the unsaturated share: the fraction of a food's fat that is the good kind. Nutrication scores that share on a 15-point scale in four bands:
| Unsaturated share of total fat | Points (of 15) | Typical residents |
|---|---|---|
| 75% or more | 15 | Olive oil, most nuts, salmon, avocado |
| 55–74% | 10 | Eggs, poultry, pork |
| 35–54% | 6 | Beef, cheese, cream, many pastries |
| Under 35% | 2 | Butter, ghee, coconut oil |
Compare this with Nutri-Score, which penalizes saturated fat by amount — gram thresholds per 100 g. That is useful, but it is half the story. Amount-based scoring cannot distinguish almonds, which are half fat but nearly all of it unsaturated, from a low-fat cookie. A ratio can. This composition blind spot is exactly why olive oil, nuts, avocado, and fatty fish historically needed hand-curated consensus exceptions in Nutrication's own engine; measuring the ratio fixes the problem structurally.
Same ratio, different food: the source discount
Here is the catch. A bag of chips fried in sunflower oil can post an unsaturated share as high as a handful of almonds. Should they score the same? The landmark evidence for unsaturated fat — Mediterranean-diet trials built on virgin olive oil, nuts, and fish — comes from intact sources, not from refined oil that has been through an industrial fryer.
So Nutrication halves the fat-quality credit when the good ratio comes from refined oil in a heavily processed food: a 15 becomes 8, a 10 becomes 5, a 6 becomes 3, and the score never drops below 2. The discount fires for ultra-processed foods, and for processed foods where a refined oil — vegetable, sunflower, canola, corn, soybean, palm, and their cousins — sits in the top three ingredients. Ingredient lists are ordered by weight, so a refined oil that high up is the dominant fat.
The exemptions are deliberate. Olive oil and avocado oil are not on the refined-oil list. Minimally processed culinary oils like extra virgin olive oil keep full credit. And a refined oil as a minor ingredient — the small pour in a tub of hummus — keeps full credit too. The rule targets one thing: fried and extruded snacks wearing a healthy fat profile they did not earn.
When fat quality isn't rated
The fat composition of an apple, a soda, or a bag of plain rice is meaningless — there is barely any fat to compose. Below 3 g of fat per 100 g, Nutrication does not rate fat quality at all. The same applies when saturated-fat data is simply missing from a product's record: the app never guesses a value. In both cases the measure is excluded and the grade is re-normalized across the measures that remain, rather than padded with a fake neutral score. The mechanics of that re-normalization are covered in how we grade.
The practical consequence: low-fat foods are neither rewarded nor punished on fat quality. They are graded on everything else — processing level, sugar, sodium, fiber, micronutrients.
Practical swaps
- Cooking fat: butter to extra virgin olive oil. This moves the fat from the bottom scoring band to the top one — the single biggest upgrade available in most kitchens.
- Snack: chips to a handful of nuts. Similar fat grams, but intact-source unsaturated fat instead of discounted refined oil.
- Dressing: bottled creamy ranch to an olive-oil vinaigrette. Same job, but the fat comes from olive oil at full credit rather than from a refined oil that earns the processed-food discount.
- Protein: trade some processed meat for oily fish. The American Heart Association's benchmark is about two servings of oily fish per week.
- Spread: if a margarine or shortening lists "partially hydrogenated" anything, put it back. Rare since 2018, but not extinct.
How Nutrication uses this
Fat quality is one of the six measures behind every grade, worth up to 15 points, and the engine cites the AHA Presidential Advisory (2017), the PREDIMED trial, and the Cochrane 2020 review as its basis. The computation is exactly what this article described: unsaturated fat is derived as total minus saturated minus trans, and the unsaturated share maps to the four bands above — 15 points at 75% or more, 10 at 55%, 6 at 35%, 2 below that.
Order of operations matters. After the band score, the refined-oil discount halves credit for ultra-processed foods and refined-oil-dominant processed foods. Then the trans-fat floor applies last and overrides everything: more than 0.2 g of trans fat per 100 g, or the phrase "partially hydrogenated" anywhere in the ingredients, caps the score at 1 point out of 15 — even a food with an otherwise perfect profile. Trace trans below that threshold skips the floor but is still subtracted when deriving unsaturated fat. Partially hydrogenated oil is also the single largest penalty in the separate ingredient-flags measure, docking 10 of its 20 points.
A perfect 15 earns the "Heart-healthy fats" highlight on the grade card; a triggered trans floor produces a "Contains trans fat" warning. And sitting above the whole algorithm is a small consensus safety net: foods with near-unanimous medical backing, like extra virgin olive oil (lifted to a 92, grade A) and walnuts (90), can be raised if the algorithm underrates them — but never lowered. It is a floor, not a ceiling. Contested foods like butter, coconut oil, and dark chocolate deliberately get no such lift; for them, the numbers speak for themselves.
What to actually do
- Stop judging foods by total fat grams. Judge them by what the fat is.
- Get most of your fat from intact sources: olive oil, nuts, seeds, avocado, fish.
- Scan ingredient lists for "partially hydrogenated" — the one phrase with no defensible amount.
- Treat a great fat ratio on a fried snack with suspicion; the app already does.
- Don't chase fat quality on foods that barely contain fat — it isn't rated there, on purpose.