Virtual makeup try-on for beauty retailers: technology and use cases

More than 60% of online beauty shoppers have decided against a purchase because they were unsure which shade would suit them, while 41% have returned a product because the shade was wrong, according to Google's research into beauty shopping behaviour. Prestige makeup sales still grew 5% across the UK and four other European markets in 2025, per Circana's B-Book 2026. The losses trace to one specific moment in the purchase, when a shopper has to judge a colour from a model's face rather than their own, under lighting and skin tones that have nothing to do with theirs.

Virtual makeup try-on gives customers a way to see that colour on their own face before they buy. Using a live camera or uploaded photo, it shows the selected shade against their own skin tone and features, giving them a much better reference than a model image alone. For products where colour plays a large part in the decision, that can remove some of the uncertainty that causes customers to hesitate or choose the wrong shade.

The preview can only help cover colour and finish accurately. Oxidation, transfer, longevity and how a formula sits on a particular skin type still depend on using the product itself.

Virtual try-on vs shade matcher, product finder, beauty filter and skin analysis

Beauty retailers put try-on, shade matching, filters and skin analysis under one label because they all use a camera. The table below separates them by what each one actually does.

A shopper might use a shade matcher to narrow the choice, try virtual try-on to see that shade on their own face, and book a consultation if they're still unsure, often within a single visit. A shade finder works before a shade is picked, recommending one from the shopper's skin tone or preferences. Virtual try-on works after, showing how that specific shade actually looks once applied.

How does virtual makeup try-on technology work?

The shopper opens the camera or uploads a photo, the system identifies the relevant parts of the face, and the selected product is placed over them. The try-on can also connect to the retailer's product catalogue, so the shade shown is tied to the product that is actually available to buy.

Capture and permission

The shopper first gives permission to use the camera or uploads a photo. The interface explains why access is needed and how the image is used. A shopper who prefers not to use the camera, or whose device does not support live try-on, can upload a photo or try the shade on a model instead.

Detecting the face

Once the camera or photo is active, the system identifies key areas of the face, such as the lips, eyelids, brows and cheeks. It maps the makeup to those areas and keeps it within the right boundaries, so lipstick stays on the lips and eyeshadow stays around the eyes. In live try-on, the makeup also follows the face as the shopper moves or changes expression.

Building the digital shade

Each shade has its own product information, including its colour, coverage and finish, such as matte, satin or shimmer. The system uses this information to render the product on the shopper's face and adjusts the result to their skin tone and the lighting captured by the camera. This means the same shade can look slightly different from one person or setting to another.

Connecting the try-on to commerce

The shade shown in the try-on is linked to the corresponding product and variant in the retailer's catalogue. When the shopper changes shades, saves a look or adds the product to their basket, those actions connect to the correct SKU, price and stock information.

How accurate is virtual makeup try-on?

Virtual makeup try-on is accurate at showing colour, placement and finish when the product data, lighting and tracking all work well. Beyond that, accuracy depends on the category and on what the shopper is actually trying to judge.

What accuracy actually measures

Where accuracy varies by category

Foundation is the hardest product to preview accurately. A lipstick sits on top of the lips as one fairly even colour. Foundation is meant to blend into the skin, so the render has to account for the shopper's own skin tone and texture, not just the product's colour. Camera settings and screen differences change how that colour looks too, so the same render can appear warmer or cooler depending on the device. Coverage and finish, how sheer or full it looks, matte or dewy, come from how the product behaves once it's actually on the skin, which a still preview can only approximate. A shade matcher, a sample or an adviser gives the shopper more to go on than the preview alone.

These tools also need to be tested properly. One built and tested mainly on a narrow set of skin tones or lighting conditions will render less accurately for anyone outside that set, sometimes lightening, greying or oversmoothing skin it wasn't built to handle. Testing across a wide range of skin tones, ages, face shapes, lighting and devices is what catches that before shoppers do.

Eight virtual makeup try-on use cases for beauty retailers

Virtual try-on can be used at several points in the beauty shopping experience, depending on what the customer is trying to decide.

1. Product pages: Customers can try different shades and finishes without leaving the product page, while keeping the price, availability and add-to-basket option in view. Retailers can then track whether people who use try-on are more likely to add the product to their basket or buy it.

2. Category pages: Try-on can also help customers compare products across different brands rather than opening each product page separately. This can be particularly useful for discovering brands or shades they may not have considered otherwise. Useful measures include the number of products tried, product-page visits and new-to-brand purchases.

3. Foundation and complexion matching: For complexion products, shade recommendations can work alongside virtual try-on. A customer answers a few questions about their skin and preferred coverage, receives a shortlist of shades and compares them on their own face. Retailers can look at shade-finder completion, repeat purchases of the same shade and customer-service queries about poor matches.

4. Full looks and bundles: Instead of trying one product at a time, customers can see how several products work together. A retailer might create a seasonal look using lipstick, blush and eyeshadow, which the customer can try as a set and then change individual products or shades. Items per order, bundle purchases and purchases across categories can show whether this leads to larger baskets.

5. Tutorials: Virtual try-on can make tutorials shoppable. As a customer follows each step, they can try the product being used and go directly to the corresponding product page. Retailers can track how many people complete the tutorial, which products they click on and how many go on to buy the full look.

6. Consultations: During an online beauty consultation, the adviser and customer can compare the same shades and products using try-on. Once they have settled on a selection, the adviser can send the products as a saved basket or direct links. Consultation-to-order rate, average order value and later purchases can help measure whether the service is leading to sales.

7. In store: Virtual try-on can also extend what customers can try in a physical store. A tablet, smart mirror or QR code can give them access to shades held in the wider catalogue rather than just what's out as counter testers. Retailers can measure assisted sales, tester use and orders placed for products outside the store's immediate stock.

8. Campaigns and merchandising: Try-on can also live inside a campaign or ad, before a shopper ever reaches the product page, letting someone try the featured products from there instead. The resulting activity can also show which shades attract the most interest. Retailers can compare campaign click-through, new-to-brand purchases and the shades being tried with the stock available.

One Journal of Marketing study followed 160,400 customers and 2,300 products at an international cosmetics retailer over 19 months, finding that sessions using AR try-on saw more browsing, more products viewed and a higher purchase rate than sessions without it, evidence of the mechanism working, not a guaranteed result for every retailer.

How should a beauty retailer implement virtual makeup try-on?

Step 1: Start with one customer decision

Start with one specific problem to solve: lip exploration, foundation uncertainty, full-look attachment, remote advice, or store throughput. A single goal gives a clear result to measure once the feature is live.

Step 2: Choose the pilot assortment carefully

Include products with reliable physical references, genuine shade and finish variation, and enough traffic behind them to produce a useful result. Pairing one high-uncertainty category, foundation, with a simpler one gives a fair comparison. Existing bestsellers convert well regardless of try-on, so a pilot built only from those tells a retailer little about the feature itself.

Step 3: Prepare the product data

Every shade needs proper product data behind it: a stable variant ID, an approved colour reference, coverage, opacity, finish, and any category-specific rules such as liner width, along with current pricing, stock and an image. This preparation determines how convincing the render looks.

Step 4: Calibrate and get sign-off

Before launch, compare the virtual result with physical swatches and the product as it looks when applied. Test it using your own catalogue rather than relying on a vendor demo, particularly with shades that are harder to reproduce accurately, such as deeper foundation shades or sheer lip tints, and across a representative range of faces. Product and brand teams should review the results alongside engineering, with a record of which shades have been approved and when they need to be checked again.

Step 5: Design the experience and its fallbacks

Explain camera permission clearly upfront, and build working photo and model fallbacks for shoppers who skip the camera. Keep price, stock and the basket action visible throughout, and make shade switching instant.

Step 6: Connect it to commerce systems

Keep the product feed, inventory and price in sync with what the shopper sees, and track exposure, renders and orders without storing raw face images to do it. Create a saved look only when the shopper takes that action deliberately.

Step 7: Check privacy and accessibility

Check data handling and retention before launch, along with whether the ICO's Children's code applies, and test the experience against accessibility standards.

Step 8: Launch as a test and keep monitoring it

Where possible, launch to a limited share of traffic first. Set up a clear route for reporting a wrong shade match or an unrealistic render, and revisit the calibration whenever a formula or asset changes.

How should retailers measure virtual try-on ROI?

The main measure should be incremental contribution per eligible session. How often customers use virtual try-on is useful to know, but the stronger measure is whether having the feature available leads to more profitable sales.

Contribution per eligible session = (net sales − cost of goods − variable fulfilment, payment, service and return costs) ÷ eligible sessions

For example, if a retailer's eligible sessions generate £50,000 in incremental net sales after costs over a month, across 20,000 sessions where try-on was available, the contribution works out to £2.50 per eligible session, whether or not the shopper actually opened the feature.

Conversion, revenue per session, product-page-to-basket rate, average order value, items per order and contribution margin can show whether try-on is changing the value of a sale. Other measures depend on how the feature is being used. For foundation, that could mean looking at repeat purchases of the same shade and refund or exchange rates. For full looks, cross-category purchases and bundle completion matter more, while consultations can be judged by how often they lead to an order. Stores can also track whether virtual try-on reduces the use of physical testers and applicators.

It is also worth looking at what happens while customers are using the feature. How reliably does it load? Are customers using the camera or choosing another option? Which products and shades are they trying, and how often do those products make it into the basket? Render errors, out-of-stock shade attempts and complaints about poor matches can help explain why the sales numbers are moving, or why they are not.

When measuring the overall effect, the comparison should include everyone who was offered a virtual try-on, not just who opened it. Customers who choose to use the feature may already be more interested in buying, so their results alone can make try-on look more effective than it is. The Journal of Marketing study cited earlier accounted for this by measuring the effect across customers exposed to AR, rather than treating usage alone as evidence of its impact. Retailers can take a similar approach by giving try-on to one group of eligible customers and comparing the results with a group that does not have access to it. Stock and promotions should remain broadly similar between the two, and the test should run long enough to see whether customers return to buy the same shade again.

The return can then be calculated against the full cost of introducing and running the feature:

Annual net value = incremental contribution + verified savings on service, testers and operations − platform, integration and running costs

This includes licensing and integration as well as shade digitisation and calibration, quality checks, privacy and accessibility reviews, catalogue maintenance and support. An in-store rollout may also bring hardware and staff training costs.

At the end of the pilot, the retailer should have a clear view of whether the additional contribution and any cost savings are enough to justify the cost of running virtual try-on.

UK privacy, advertising and accessibility considerations

This is practical guidance rather than legal advice, so retailers should confirm the requirements for their own setup.

Under UK GDPR, the way a facial image is processed and the reason for using it determine how it is classified. An image used to position virtual makeup is treated differently from one processed to recognise or identify a person, which can fall under special category biometric data. Before the camera opens, retailers should explain how the image will be used, whether it will be stored, how long it will be kept and when it will be deleted.

What customers see on screen should give them a fair view of the product they are considering. ASA and CAP advertising rules apply to cosmetics shown through virtual try-on, so the render should stay focused on the product itself, with digital shades checked against their physical versions and effects such as skin smoothing or facial reshaping kept separate from the result.

Customers may use virtual try-on in different ways, and the experience should remain accessible across them. Following WCAG 2.2, shade options and controls should work with a keyboard, with shade names or other cues alongside colour to make each option clear. Photo, model or other alternatives to live camera try-on should carry the same product information customers need to compare shades and make a choice.

Bringing virtual makeup try-on to retail with Fynd GlamAR

Fynd GlamAR supports live camera and photo-based makeup try-on across websites, apps and stores. It covers lipstick, foundation, eyeshadow, blush and other categories, with individual and full-look try-on. SDKs and APIs connect it to existing commerce systems, while catalogue onboarding and analytics are included. GlamAR does not store personal data or try-on images, though retailers should confirm the terms for their own implementation.

A focused pilot is a good place to start. Choose a small part of the catalogue where colour is difficult to judge online, calibrate the digital versions against the real ones and test them across a representative range of faces. Conversion, order value, repeat purchases and returns can then help decide whether a wider rollout is worthwhile.

The preview is useful for judging colour and finish before buying. Wear, transfer, oxidation and how a formula feels on the skin still need to be experienced in person. Keeping that distinction clear gives virtual try-on a practical role in helping people choose makeup online with more confidence.

Want to see how this works for your own catalogue?

Book a demo