AI skin analysis apps or a skin consultation?

01 / The trend

What AI skin analysis apps actually do.

Most skin analysis apps work from a photograph. Software looks at the image and gives scores or labels for what it can see: lines, spots, redness, shine, visible pores. Many then estimate a “skin age” and recommend products, often from the brand or retailer behind the app.

That last part is worth noticing. In Dermatology Times, dermatologist Zoe Draelos says facial appearance apps should not be relied on for accurate age assessment, and describes them as a targeted route to skincare product sales.[4] That does not make every app useless. It means the recommendation at the end is not neutral, and it is worth separating what the app saw from what it wants you to buy.

What the trend says

The app “knows” your skin type, your skin age and exactly which products you need.

What is actually known

An app reads visible features in one image. Its output depends on the photo, and its product list depends on who made it.

02 / The limits of a selfie

What a photo cannot show.

Light alone can change the result. In tests described by Dr Draelos, people appeared about three years younger under incandescent light than under fluorescent or LED light, and the apps did not account for that difference.[4] Camera quality, angle, distance, make-up and whether you have just washed your face all play a part. Even dermatologists ask for photos taken in good light with make-up removed, and suggest avoiding a hot shower, rubbing or skincare products beforehand, because these change how skin looks.[5]

Then there is everything a camera cannot see at all:

• how your skin feels: tight after cleansing, stinging with a product, comfortable by midday or not;
• what you use now, how often, and in what combination;
• what you have tried before, and how your skin reacted;
• recent changes in season, stress, medication or routine.

Two people with the same score can need very different things. Dry and dehydrated skin can look alike in a photo, for example, while needing a different approach; dry or dehydrated skin? explains the difference, and how to understand what your skin may actually need shows how much depends on observation over time.

03 / Who the software was trained on

Skin tone, training data and accuracy.

An AI tool learns from the images it was trained on, and it can struggle with images that look different. A 2022 study reported by Stanford Medicine found much poorer performance of dermatology algorithms on images of Black and brown skin; algorithms trained mostly on lighter skin performed well on lighter skin.[3] Researchers at Memorial Sloan Kettering also found that stray hairs or pen marks in a photo reduced an AI tool’s accuracy, and that on image types it had not been trained on, its performance was close to chance.[2]

Those studies looked at medical algorithms rather than beauty apps, and consumer apps rarely publish how they were tested. That is exactly the point: if you do not know what an app was trained on, you do not know how well it reads your skin.

04 / A firm line

Moles are not an app question.

Some apps go further and claim to assess moles for skin cancer. The British Association of Dermatologists has warned that AI skin cancer apps are endangering the public, that the published evidence that AI can diagnose skin cancer safely and effectively is weak, and that many such apps lack the regulatory approval a medical device needs.[1] In tests of commercial apps described by Memorial Sloan Kettering, the apps identified melanoma correctly only 59% of the time on average, and the researchers warned of a false sense of security.[2]

05 / The other option

What a skin consultation adds.

A consultation with Merve Akkaya is a focused conversation, online or in person in Istanbul, by appointment. Rather than scoring one photo, it looks at four parts of the picture:

Your skin now

Its present condition, visible concerns, comfort, sensitivity and recent changes.

Your routine

Products, active ingredients, combinations, frequency, and anything that may be adding unnecessary complexity.

Your history

Previous professional care, reactions, what has helped and what has not made sense for you.

Your priorities

What you genuinely want to improve, and what kind of plan you can realistically maintain.

The aim is not the greatest number of products or treatments. It is a clearer next step: what to maintain, reconsider, simplify or explore further. Sometimes that means adding less. A consultation focuses on skincare and is not a medical diagnosis; where a symptom needs a doctor, that is the more suitable next step. What a useful skin consultation should leave you with goes into more detail.

06 / If you like your app

How to use a skin app sensibly.

1. Keep the conditions the same.

   Same place, same light, same time of day, bare skin. The tips dermatologists give for appointment photos apply here: good light, no make-up and no hot shower, rubbing or products just before.[5] Otherwise you are measuring the lighting, not your skin.

2. Watch the trend, not the number.

   A single score means little. Photos taken the same way over weeks can show whether something is changing, much like keeping a skin diary.

3. Pause before the basket.

   If the app recommends products, check what you already own first. A new product added on top of a full routine can cause more problems than it solves; when more is not better explains why.

4. Turn scores into questions.

   “The app says my redness is high” is a good starting point for a conversation. It is not a diagnosis, and persistent redness deserves a proper look.

A few common questions

Frequently asked questions.