Virtual Makeup Try-On Adoption: What L'Oréal's 100 Million Sessions Mean for Smaller Brands

Virtual makeup try-on adoption stopped being an experiment somewhere around 2023. The clearest public marker comes from the largest player in the category: L'Oréal registered more than 100 million virtual try-on sessions in 2023, against 40 million in 2022, with usage that started during the pandemic and kept growing afterwards, moving from lipstick into other categories (Nicolas Hieronimus, CEO of L'Oréal, annual results presentation, February 9, 2024). That is a third-party number, reported by L'Oréal about L'Oréal — it is not a MaIA result and it is not a promise of what your brand will see.

Still, it is the most useful data point available to a CMO or e-commerce lead deciding whether try-on deserves a line in next quarter's roadmap. Not because of the size of the number, but because of what kind of number it is.

What this number actually says — and what it does not

Sessions are an adoption metric. More than 100 million sessions in 2023 versus 40 million in 2022 (Nicolas Hieronimus, CEO of L'Oréal, annual results presentation, February 9, 2024) tells you that shoppers voluntarily chose to try a shade on their own face, at scale, year after year. It does not tell you how many of those sessions ended in a purchase, an add-to-cart or a lower return rate. Anyone converting that figure into a conversion claim is inventing something.

Why does adoption still matter? Because adoption is the part of the equation a brand cannot buy. Features that shoppers ignore are the default outcome of most e-commerce projects. A feature used voluntarily, repeatedly, and then extended from lipstick into face and eye categories is behaving like a habit — and habits survive redesigns, budget cycles and new traffic sources.

The fact that usage spread beyond lipstick

The category detail is the quiet insight. Lipstick is the easy entry point: high shade count, low commitment, obvious visual payoff. When the same behaviour spreads to foundation, blush and eyes, it means the underlying need was never "lipstick". It was "I cannot tell if this colour works on me from a swatch on a white background".

Why "is this shade for me?" is the mechanic behind the habit

Every colour cosmetic sold online carries the same friction: the shopper has a face, the page has a circle of pigment, and nothing connects the two. That doubt sits directly on the path to purchase. It produces abandoned carts, shade-guessing, and returns that cost twice — once in logistics, once in trust.

Virtual try-on works because it removes that specific question from the path. Not because it is technically impressive. Which is exactly why the mechanic is portable: it does not depend on a global marketing budget, a flagship store or a proprietary device. It depends on showing the shopper the product on their own face before they decide.

That is what MaIA Try-On does in its photo-based form: the customer sends a photo, the lipstick or foundation is rendered on it, and the image comes back — with palettes calibrated to your real products, not generic colours that look nothing like what ships in the box.

Virtual makeup try-on adoption without a global budget

A smaller brand does not enter this game by rebuilding L'Oréal's stack. It enters by placing the same moment of proof in the three places where shade doubt already appears in its own funnel.

On the product page

The shopper is already on the PDP, already interested, already stuck on shade. A photo-based try-on turns the shade selector into something testable instead of something guessed. This is the highest-intent placement and usually the first one to instrument.

In the shade quiz

Most beauty quizzes end with a recommendation the shopper has to trust blindly. Pair the quiz with a photo: read the beauty profile with the MaIA Scan API — skin tone and undertone, skin type and condition, hair attributes, each field with a confidence score — then show the recommended shade rendered on the same photo. The recommendation stops being an assertion and becomes a preview.

On WhatsApp

In Brazil, a large share of beauty conversations happens in chat, often with a human or a bot answering "which shade should I buy?". Because the API works from a photo and returns an image, it fits a conversation natively: the shopper sends a selfie, receives the product rendered on it, and replies. No app download, no store visit.

Photo-based, live camera mode, and why the distinction matters

MaIA Try-On exists in two forms, and choosing between them is a planning decision, not a marketing one. The API is photo-based: send a photo, get an image back, with lipstick and foundation. The browser SDK, in early access since September 2026 and granted per account, runs in live camera mode — the customer's camera rendered locally in the browser, nothing uploaded — or in photo mode, covers a wider set of products including blush, eyes and complexion plus a combined Full Look, accepts any #RRGGBB colour, and is self-hosted by the brand. We wrote a full comparison of the two approaches in photo-based virtual try-on vs AR try-on.

Start where your traffic already is. A photo-based flow on the PDP, in the quiz and in WhatsApp covers most of the shade-doubt surface a mid-sized brand actually has.

How you find out what it is worth to you

L'Oréal's sessions figure proves the behaviour exists. It says nothing about your numbers. Your conversion, add-to-cart and return impact depend on your catalog — how many shades, how similar they are, how well photographed — on your traffic mix, and on the channel where try-on sits. Those variables move results enough that borrowing someone else's percentage would be dishonest. We measure it inside the project, against your baseline, on your own funnel.

Ready to test it on your catalog? Create an account and an API key in the MaIA API documentation, or talk to the B4A team about where try-on fits in your funnel.