Virtual Try-On Conversion Rate: What the Perfect Corp Report Means for Beauty Brands

Every beauty e-commerce team eventually asks the same question in a budget meeting: what does virtual try-on actually do to the virtual try-on conversion rate? A widely circulated figure says brands reported a 2.5× increase in sales conversion after implementing virtual try-on (Perfect Corp AI trends report, via Retail Dive, Sep 3, 2024). That number is worth discussing — but it is worth discussing carefully, because of who published it and because no third-party benchmark is a promise about your catalog.

Where the number comes from, and what it is not

The figure appears in an AI trends report published by a try-on vendor and covered by Retail Dive. The same report states that 38.4% of surveyed executives pointed to virtual fitting rooms and virtual try-ons as the most effective use of AI and AR to bring the digital experience closer to the physical one (Perfect Corp AI trends report, via Retail Dive, Sep 3, 2024).

Two caveats, stated plainly:

• This is a third-party number, not a MaIA result. We did not produce it, we did not audit the methodology behind it, and we are not presenting it as what MaIA Try-On will do for you.
• It comes from a supplier's own report. A company that sells try-on technology publishing research favourable to try-on technology is normal in B2B, and it is also a reason to treat the figure as directional rather than as a forecast.

So why cite it at all? Because the direction is useful, and because the mechanism implied by it is something you can reason about, test and measure in your own store.

The mechanism: removing "is this shade for me?" from the purchase path

A 2.5× lift is an outcome. The cause is much less glamorous: in colour cosmetics, the main friction is not price, shipping or trust in the brand. It is a single unresolved question — is this shade for me?

That question is what makes a shopper open twelve product pages, zoom into swatch photos taken on someone else's skin, read reviews looking for "I'm medium with yellow undertone" and then close the tab. It is also what drives returns and exchanges after purchase, and what makes the first purchase of a foundation feel risky enough to postpone indefinitely.

Anything that answers that question earlier in the journey — with the shopper's own face, not a model's — removes a decision the shopper was not equipped to make. That is the engine behind the reported 2.5× lift in sales conversion (Perfect Corp AI trends report, via Retail Dive, Sep 3, 2024), and it is not exclusive to any one rendering technology.

Why photo-based try-on triggers the same mechanism

The MaIA Try-On API is photo-based: the shopper sends a photo, the product — lipstick or foundation — is rendered on that photo using a palette calibrated to your real products, and the image comes back. There is no live camera stream in the API, and we do not describe it as AR.

That difference matters less than teams assume, because the shopper's question is answered either way. A rendered photo of her face wearing your shade 220 settles the doubt as effectively as a moving preview does — and in some channels it works better, because a photo can be sent, saved, compared side by side and shown to a friend. If you want the full comparison of the two approaches and when each one fits, we wrote it up in photo-based virtual try-on vs AR try-on.

Three places to apply it

The product page

The classic placement. A shopper lands on a foundation PDP from paid media, uploads a selfie, sees the shade on her own skin and either adds to cart or switches to a neighbouring shade that fits better. The metric to watch is not only conversion: look at add-to-cart rate on pages with try-on versus without, and at shade-switching behaviour before checkout.

The quiz

A shade-finder quiz that ends with a recommendation is good. A quiz that ends with a recommendation rendered on the shopper's face is a different experience. Pairing analysis with try-on is the natural combination here: MaIA Scan returns a beauty profile from a selfie — skin tone and undertone, skin type and condition, hair attributes, with a confidence per field — and try-on turns that recommendation into something the shopper can see. These are visible attributes of a photo, not a diagnosis, and they are not a substitute for a professional.

WhatsApp and conversational channels

In Brazil, a large share of beauty discovery and purchase happens in conversation. A photo-based flow fits a chat thread natively: the customer sends a selfie, receives her shade applied, replies with a question, receives an alternative. Connected to MaIA Beauty Advisor, the same profile feeds recommendations drawn from your own catalog instead of generic advice.

One boundary worth stating: the eleven other products and the Full Look — blush, eyeshadow, mascara, eyeliner, brows, bronzer, contour, highlighter, powder, concealer, lip liner — are available in the browser SDK, in early access since September 2026, rendered locally on the shopper's device in live camera mode or photo mode. The API renders lipstick and foundation.

How to measure your own number

Do not adopt someone else's benchmark as a target. Set up the measurement instead:

1. Choose one category and one channel — for example, foundation on the PDP.
2. Split traffic, with and without try-on.
3. Track conversion, add-to-cart, shade switches, session depth and, after 30–60 days, exchange and return rates.
4. Only then decide where to expand.

Conversion depends on your catalog depth, your traffic quality and the channel where the experience lives. A brand with six foundation shades and a brand with forty will not see the same effect. That is exactly what we measure together during the project, on your data, instead of quoting a figure from someone else's report.

Ready to test it with your own shades? Create an account and get an API key, or talk to our team about a try-on and advisor project built on your catalog.