The Science Behind Personalized Makeup Recommendations

Why the Right Shade Is a Physics Problem, Not Just a Beauty Problem

Quick Answer: Personalized makeup recommendations work because skin color and undertone are measurable, physical properties, not aesthetic guesses. Skin tone comes from a mix of pigments (mainly melanin, hemoglobin, and carotenoids) that scatter and absorb light in specific, quantifiable ways.[1][2] Foundation shade matching and color analysis draw on this science, along with newer, more inclusive measurement tools like the Monk Skin Tone Scale, to reduce the guesswork that has historically led to mismatched shades and washed-out color choices.[3][4] AI tools can now analyze these light patterns from a photo, but accuracy still depends on lighting, camera quality, and how much personal context a person provides.

Anyone who has stood at a drugstore shelf holding two nearly identical foundation bottles knows the frustration. One looks perfect in the store's lighting and turns orange by lunchtime. The other seems close enough, until it sits on top of the skin instead of blending into it. For decades, the beauty industry treated this as a problem of trial and error. It isn't. It's a problem of light, pigment, and measurement, and there's real science behind why some matches work and others don't.

What Actually Determines Skin Tone and Undertone

Human skin color comes from a small set of pigments, called chromophores, that absorb and reflect light. The three primary contributors are melanin, hemoglobin, and carotenoids. Melanin, produced in the epidermis, is the dominant factor in how light or deep a skin tone appears. Hemoglobin, circulating in blood vessels beneath the skin, adds red and pink tones, while carotenoids from diet contribute yellow and orange hues.[1][2] Research on skin optics confirms that these chromophores interact with light in measurable, wavelength-specific ways, which is why spectrophotometers and colorimeters can quantify skin color with far more precision than the human eye alone.[5]

This is also where the distinction between skin tone and undertone becomes useful. Skin tone refers to depth, how light or dark the skin appears, and is driven mostly by melanin concentration. Undertone refers to the underlying warmth or coolness of that color, shaped by the balance between hemoglobin and carotenoids. The two are independent of each other. A deep complexion can lean warm or cool, and so can a fair one. This is part of why "warm equals light" or "cool equals dark" assumptions, common in casual color-matching advice, don't hold up scientifically.

Why the Fitzpatrick Scale Wasn't Built for This

For years, cosmetics and tech companies borrowed the Fitzpatrick scale, a six-category system, to describe skin tone. The problem is that the Fitzpatrick scale wasn't designed for this purpose. It was created in 1975 to estimate how skin responds to ultraviolet light and predict skin cancer risk, not to describe cosmetic color.[6] Independent research has found that Fitzpatrick categories correlate more closely with self-reported race than with objectively measured skin color, and that the scale underrepresents the range of darker skin tones.[6]

In response, Google researchers partnered with Harvard sociologist Dr. Ellis Monk to develop the Monk Skin Tone (MST) Scale, a 10-shade system built specifically to represent skin tone diversity more accurately.[3] A nationally representative study comparing the Fitzpatrick scale, Rihanna's Fenty Beauty 40-shade palette, and the Monk scale found that people, especially those from historically underrepresented groups, perceived the Fitzpatrick scale as notably less inclusive than the other two.[4] The MST scale is now open-sourced and used across computer vision and beauty tech to reduce the kind of bias that left many skin tones poorly served by earlier shade-matching systems.

How AI-Based Shade Matching Actually Works

Modern shade-finding tools, whether built into a retailer's website or a broader beauty platform, generally work by analyzing a photo or short video, then estimating skin tone and undertone from the light values captured in the image. That data is then cross-referenced against a database of existing foundation shades, sometimes spanning tens of thousands of products, to suggest the closest matches across brands and price points.

This approach is genuinely useful. It removes some of the subjectivity of relying on in-store lighting and staff judgment, and it can compare a person's skin against a far larger shade library than any single counter could stock. But it has real constraints. Camera sensors, screen calibration, and ambient lighting all affect the color values a photo captures, and inconsistent lighting remains one of the most common reasons a digital shade match doesn't hold up once someone gets the product home. Fluorescent light adds a cool, blue-green cast, while incandescent bulbs skew warm and yellow, and both can distort how undertone is read from an image. This is a known limitation of image-based color analysis generally, not a flaw unique to any one tool.

Beyond Foundation: How Color Theory Extends to the Rest of the Face

The same undertone logic that governs foundation matching also shapes recommendations for blush, lipstick, and eyeshadow. Color theory holds that shades harmonizing with a person's underlying pigmentation tend to read as more flattering, while clashing shades can make skin look sallow, ashy, or overly flushed. This is the logic behind personal color analysis systems that sort people into categories like "warm autumn" or "cool winter" based on undertone, depth, and contrast.

It's worth being precise about what's established here and what isn't. The pigment science behind undertone, melanin, hemoglobin, and carotenoid ratios, is well documented in dermatology and optics research. The specific four-season or twelve-season color categories used in commercial color analysis are a practical framework built on that science, not a peer-reviewed clinical system, and different consultants can sort the same person into different seasons. The usefulness of color analysis lies in the underlying principle: shades that echo a person's natural pigment balance tend to look more cohesive on skin, and that principle is worth understanding even if the exact category labels are more art than science.

Where Skin Health Intersects With Color Choice

Color accuracy is only part of a good makeup recommendation. What sits on the skin for hours also matters. The American Academy of Dermatology notes that ingredients labeled "non-comedogenic," "oil-free," or "won't clog pores" are less likely to trigger breakouts, and that people prone to acne can wear makeup safely as long as they choose formulas suited to their skin type and remove them properly at the end of the day.[7] A shade can be a perfect color match and still be the wrong product if it doesn't suit someone's skin type or sensitivities. Good personalized recommendations account for both dimensions at once, rather than treating color and formulation as separate questions.

What This Means for Your Beauty Routine

If you're trying to find a genuinely well-matched shade, treat any single tool, human or AI, as a starting point rather than a final answer. Test recommended shades in natural daylight near a window rather than under indoor lighting, since both fluorescent and incandescent bulbs distort undertone. Check a potential match against your jawline and neck, not just the back of your hand, since color and texture can differ across those areas.

It also helps to describe your skin in more than one dimension when asking for advice, whether from a person or an AI tool. Depth, undertone, and even how your skin responds to certain formulas are three different pieces of information, and providing all three tends to produce a more useful recommendation than a single photo alone. New beauty platforms, including FaceEcho, are building tools that combine this kind of shade and undertone analysis with personal preferences and beauty-expert input, rather than relying on image analysis by itself.

Finally, remember that a color match is not the same as a skin-safe match. A shade can look right and still irritate sensitive skin, so cross-checking formulation against your skin type remains a separate, necessary step.

Frequently Asked Questions

1. What determines someone's natural skin undertone?Undertone comes primarily from the balance between hemoglobin (which adds red and pink) and carotenoids (which add yellow and orange) beneath the skin, independent of how light or deep the skin's overall tone is.[1][2]

2. Is the Fitzpatrick scale accurate for matching makeup shades?Not reliably. It was designed to assess UV sensitivity and skin cancer risk, not to describe cosmetic color, and research shows it correlates more with self-reported race than with measured skin tone.[6]

3. What is the Monk Skin Tone Scale?It's a 10-shade, open-source scale developed by Google and Harvard sociologist Dr. Ellis Monk to represent skin tone more inclusively than older scales, particularly for people with darker skin.[3]

4. Can AI accurately determine my foundation shade from a photo?AI tools can estimate skin tone and undertone from an image and match it against large shade databases, but accuracy is affected by lighting conditions, camera quality, and screen calibration.

5. Is personal color analysis (seasonal color analysis) scientifically proven?The pigment science behind undertone is well established, but the specific "season" categories used in color analysis are a practical framework rather than a validated clinical system, and results can vary between consultants.

6. Why does my foundation look different at home than it did in the store?Lighting is the most common cause. Fluorescent lighting adds a cool cast while incandescent lighting adds a warm one, both of which can distort perceived undertone.

7. Can two people with the same skin tone need different foundation shades?Yes. Skin tone (depth) and undertone (warmth or coolness) are independent variables, so two people with similar depth can still need very different shades.

8. Does makeup ingredient choice matter as much as color match?Yes. The American Academy of Dermatology recommends choosing non-comedogenic, oil-free formulas for acne-prone skin, since the wrong ingredients can cause breakouts regardless of how well the color matches.[7]

9. What's the best way to test a foundation shade before buying?Apply it along the jawline and check it in natural daylight, since indoor lighting can misrepresent both depth and undertone.

10. Do AI shade-matching tools work equally well across all skin tones?Not always. Tools built on more inclusive scales like the Monk Skin Tone Scale tend to perform better across a wider range of skin tones than earlier systems built on the Fitzpatrick scale.[4]

References

[1] Springer Nature. Skin Color and Pigmentation. In: Handbook of Cosmetic Science and Technology reference works. Official publisher record.

[2] ScienceDirect / Elsevier. Normal and abnormal skin color. Annales de Dermatologie et de Vénéréologie. Peer-reviewed reference.

[3] Google Research. Monk Skin Tone Scale. Official Google AI documentation, developed in partnership with Dr. Ellis Monk (Harvard University). https://skintone.google

[4] Heldreth CM, Monk EP, Clark AT, et al. Which Skin Tone Measures Are the Most Inclusive? An Investigation of Skin Tone Measures for Artificial Intelligence. ACM Journal on Responsible Computing. 2024. DOI: 10.1145/3632120.

[5] PubMed. Blood stasis contributions to the perception of skin pigmentation. Skin Research and Technology. PMID: 15065897.

[6] Monk Skin Tone Scale documentation citing IEEE research on Fitzpatrick scale limitations in computer vision applications.

[7] American Academy of Dermatology Association. Makeup Tips for Acne-Prone Skin. Official guidance, aad.org.