the nerdy bit, in full
How we check the science
Glowcast does two separate things. The app gives each meal you log a Glow Score, weighted for your skin. This website grades the research on individual foods. The grades describe what studies in groups of people found. Your score describes your meal. Neither is a diagnosis, and neither can tell you how your skin will respond.
What the Glow Score looks at
When you log a meal, an AI model identifies the foods in it and rates each one against eight dimensions that research links to skin. The app then combines those ratings into one Glow Score for the meal, weighted for your profile.
- Hydration. Water content, electrolytes, and whether a food or drink tends to dehydrate.
- Anti-inflammatory. The balance of omega-3 to omega-6 fats, polyphenols, and other parts of a food linked to inflammation.
- Antioxidant. Vitamins C and E, carotenoids, selenium and similar compounds.
- Glycemic impact. How quickly a food raises blood sugar, and how much fibre and added sugar it carries.
- Nutrient density. Vitamins and minerals that skin uses for repair and renewal, relative to calories.
- Collagen support. Nutrients the body uses to make collagen, such as vitamin C, and factors that work against glycation.
- Gut-skin axis. Fibre that feeds gut bacteria, fermented foods, and ingredients that may disrupt the gut barrier.
- Hormonal impact. How strongly a food raises insulin and IGF-1, the growth-factor signal studied in dairy research.
What personalises your score
The same meal can score differently for two people, because the eight dimensions are weighted for your profile rather than averaged.
- Skin type. Normal, oily, dry or combination. For oily skin, glycemic and hormonal impact count for more. For dry skin, hydration counts for more.
- Sensitivity. Resilient, reactive or seasonal. For reactive skin, the anti-inflammatory and gut-skin dimensions count for more.
- Skin goal. Clear, brighten, calm or longevity. A brightening goal gives antioxidants more weight; a calming goal gives inflammation more weight.
- Age range and skin concerns. For example, choosing dullness as a concern gives antioxidants and hydration more weight, and choosing dryness gives hydration more weight.
- Season. Weights shift a little with the season, for example towards hydration in winter.
- Food flags. A small set of foods carry a penalty or a bonus for specific profiles, such as sugary drinks for oily skin or cured meats for reactive skin.
Over time, the app lines up your scores with how your skin looks, so you can see which patterns seem to matter for you. It shows correlation. It does not prove that a food caused a change, and skin often takes weeks to reflect a change in diet.
What the Glow Score is not
The Glow Score is an estimate built from research averages and your profile. It is not medical advice and not a diagnosis. Genetics, hormones, sleep, stress, skincare and medication all affect skin, and people differ a great deal in how they respond to the same food. If you have a skin condition, talk to a dermatologist.
How we grade the evidence
Our grades have two independent axes: human evidence and mechanism. Every article has a research section at the bottom that shows both grades for each food, the studies behind them, and the date they were checked. Some older pages still show the grade on a four-step meter.
Axis 1 — human evidence
| Grade | Definition |
|---|---|
| Strong | Two or more randomised controlled trials in people that measured a skin outcome agree, or a meta-analysis of such trials does. |
| Moderate | One randomised trial with 100 or more participants that measured a skin outcome, or observational studies in two or more separate populations that agree, with no trial contradicting them. |
| Limited | A single observational study, trials with fewer than 100 participants, or studies that disagree. |
| Too little (researchers call this insufficient evidence) | No study in people has measured a skin outcome for this food. |
Moderate means pooled studies in two or more separate populations agree and no randomised trial contradicts them; single studies that disagree are named in the research but don't lower the grade on their own.
Axis 2 — mechanistic plausibility
| Grade | Definition |
|---|---|
| Supported | The proposed pathway (for example raised insulin or IGF-1) has been measured in people after eating this food. |
| Plausible | The pathway has been shown in cell or animal studies, or in people for a closely related food or nutrient. |
| Speculative | The pathway is inferred by analogy, with nothing measured directly. |
Grading rules
- Grade the specific claim "eating <food> affects <skin outcome>", not the nutrient in general.
- When the conditions for two grades both match, use the lower grade.
- Mechanism never raises the human-evidence grade.
- Every grade is backed by a line in the page's citation log naming the studies it rests on.
Each page also shows the date its evidence was last checked.
How we choose sources
Tier 1 is peer-reviewed research, systematic reviews, clinical guidelines and government data. Tier 2 is major publications. Tier 3 is reputable industry sources. Health claims rest on tier 1 wherever tier 1 exists.
Every figure on a page built or rebuilt since September 2026 is checked against the sentence in its source that supports it, and each page keeps a citation log of those sentences. A figure we cannot find in its source is cut. Who builds Glowcast, what it is not, and the corrections log are on the About page.