10 August 2026 · 5 min read

Virtual try-on explained: what retailers need to know

Discover how virtual try-on technology boosts shopper confidence, reduces returns, and enhances purchasing decisions for retailers.

Virtual try-on explained: what retailers need to know

Virtual try-on (VTO) is a technology that lets shoppers see how a garment, accessory, or cosmetic product looks on their own body or face before buying, using a photo, live camera feed, or 3D avatar rather than a physical fitting room. It is not the same as virtual reality: VTO typically runs in a browser or app without a headset, overlaying or synthesising product imagery onto the shopper’s image in seconds. For retailers, the commercial case is direct: higher purchase confidence, fewer fit-driven returns, and stronger conversion. For shoppers, it answers the one question a product page cannot: “Will this actually work on me?”

The core benefits at a glance:

Visual confidence: shoppers see the product on their own proportions, not a stock model.
Personalised sizing: AI-driven measurement tools map body dimensions to specific garments.
Reduced returns: fewer purchases made on guesswork means fewer items sent back.
Sustainability: lower return volumes cut reverse logistics emissions and packaging waste.
Engagement and social sharing: try-on results are shareable, extending organic reach.

Key takeaways

Virtual try-on technology increases purchase confidence and reduces fit-driven returns by letting shoppers see garments on their own body before buying, and the competitive advantage now lies in digital asset quality and experience rather than proprietary rendering technology.

Point Details VTO definition Shoppers see products on their own body via photo, camera, or avatar, without a physical fitting room. Vendor-reported uplifts AR try-on has been associated with ~20% sales increases and up to 64% return reductions in vendor interviews. Hybrid systems perform best Combining AR overlays, AI sizing, and 3D simulation delivers the most practical results across mixed catalogues. GDPR applies Shopper photos and body measurements are personal data under UK GDPR; a DPA and possible DPIA are required. Competitive advantage Infrastructure is commoditising; brands win by investing in digital assets, CRM integration, and experience quality.

How virtual try-on systems actually work
The concrete benefits for shoppers and retailers
Which VTO technology fits your product category?
Challenges and limitations you need to plan for
Where virtual try-on is being used in UK e-commerce
How to implement virtual try-on: from pilot to rollout
Why the competitive advantage has shifted away from the technology itself
Where virtual try-on is heading next
Making virtual try-on accessible to every shopper
A practical perspective on where to focus your effort
Try virtual try-on with Garmcheck
Sources

How virtual try-on systems actually work

Every VTO system, regardless of vendor, processes the same sequence of inputs and outputs. Understanding that pipeline helps you evaluate trade-offs before committing to an integration.

Inputs the system needs

Three data types feed every pipeline: a customer image (photo upload or live camera), a product image (usually a flat lay or on-model shot), and product metadata (measurements, fabric weight, colour variants). The richer the product data, the more accurate the output.

The processing steps

Landmark detection: the system identifies key body points (shoulders, waist, hips, inseam) or facial landmarks (eyes, nose, jaw) from the customer image.
Body or face modelling: those landmarks generate a 2D skeleton or a 3D mesh that represents the shopper’s proportions.
Garment mapping: the product image is warped, scaled, and aligned to the body model, accounting for garment geometry where data allows.
Rendering or simulation: the system produces a final image, either by compositing layers (AR overlay), running an image-to-image diffusion model to synthesise the garment onto the body while preserving fabric detail, or by simulating cloth physics on a 3D avatar.
Output delivery: a still image, an animated preview, or a real-time AR mirror is returned to the shopper’s screen.

Three distinct rendering approaches

AR overlays: fast, lightweight, well-suited to beauty, eyewear, and accessories where precise body fit is less critical.
Image-to-image generative pipelines: use diffusion models to produce photorealistic on-body images; strong garment fidelity, but computationally heavier.
3D avatar and physics simulation: the most realistic for structured garments; highest content cost and longest render time.

Virtual fitting rooms that combine all three layers in a hybrid model tend to deliver the most practical results for merchants, because no single approach handles every product category well.

Pro Tip: For the best output quality, ask shoppers to photograph themselves against a plain, well-lit background in form-fitting clothing. Poor lighting and busy backgrounds are the single most common cause of inaccurate landmark detection, regardless of which vendor you use.

The concrete benefits for shoppers and retailers

The commercial case for VTO rests on three levers: conversion, returns, and engagement. Each has supporting evidence, though the strongest figures come from vendor interviews rather than large-scale independent studies, so they should be read as indicative rather than guaranteed.

Conversion and returns

Vendor-reported data cited by Shopify’s AR try-on guide suggests AR try-on implementations have been associated with sales uplifts of around 20% and return-rate reductions of up to 64% for some merchants. Those figures come from vendor interviews with companies including Perfitly, so treat them as directional. What is consistent across multiple sources is the direction: try-on reduces the uncertainty that causes both abandoned carts and post-purchase regret.

Engagement, personalisation, and sustainability

Beyond conversion, VTO drives personalisation and social sharing while generating richer data on which products shoppers engage with most. A shopper who tries on ten items and buys two is giving you signal that a standard browse session never would. On sustainability, fewer returned shipments means less packaging, fewer courier journeys, and lower carbon output per sale.

Benefit What it affects Evidence basis Conversion uplift Add-to-cart and checkout rate Vendor-reported ~20% sales increase Return reduction Reverse logistics cost Vendor-reported up to 64% reduction Time on site Engagement and SEO signals Vendor-reported increases Social sharing Organic reach and brand awareness Industry observation Sustainability Emissions and packaging waste Qualitative (fewer return shipments)

One-line retailer outcomes worth noting

Footwear brands using 3D sizing tools have reported measurable drops in “wrong size” return reasons.
Beauty retailers using face-AR for lipstick and eyeshadow see higher add-to-cart rates on colour products where shade uncertainty is the main barrier.
Apparel brands that generate on-model imagery at scale using production try-on pipelines report that the content itself lifts conversion across all shoppers, not just those who interact with the try-on widget.

Which VTO technology fits your product category?

Not every approach suits every product. The right choice depends on the garment type, the realism required, your content budget, and your shoppers’ devices.

Face AR Best for beauty (lipstick, eyeshadow, blush), eyewear, and jewellery worn near the face. Runs in real time via a device camera, requires no body data, and works on most modern smartphones. Vendors such as Banuba and Camweara specialise in this layer, with Banuba focused on beauty AR and Camweara on eyewear and jewellery overlays.

Eyewear overlays A subset of face AR, but worth separating because frame geometry and lens tint require precise 3D modelling. Wanna (now part of Snap) offers eyewear and footwear AR with strong 3D asset pipelines. Accuracy depends heavily on the quality of the 3D frame model supplied by the brand.

Photo-based image-to-image pipelines The shopper uploads a photo; the system synthesises the garment onto their body using a diffusion model. This approach suits full-body apparel and does not require a live camera. Swan.ai and Style.me operate in this space, generating on-body images from a single uploaded photo. OnYou takes a similar approach, focusing on photorealistic garment rendering from customer selfies. Garmcheck uses this pipeline to produce photorealistic results in under ten seconds from a single front-facing photo, with size recommendations derived from eight body measurements.

3D avatars and physics simulation The shopper builds or imports a body avatar; garments are simulated with cloth physics. Highest realism for structured pieces (tailored jackets, denim), but requires 3D garment assets and longer render times. Perfitly and Fitnonce both work in this space, with Perfitly emphasising body-shape accuracy and Fitnonce targeting fit prediction for apparel.

In-store smart mirrors Physical screens with embedded cameras that run AR or avatar try-on in a retail environment. High engagement, but significant hardware cost and maintenance overhead.

Technology type Best product categories Key trade-off Face AR Beauty, eyewear, jewellery Fast and device-friendly; limited to face/neck area Photo-based generative Full-body apparel, outerwear Photorealistic; requires quality photo input 3D avatar simulation Structured garments, tailoring Most accurate fit; highest content cost In-store smart mirror Any category High engagement; hardware investment required

Hybrid systems that layer AR overlays with AI sizing and 3D simulation tend to deliver the most practical results across mixed product catalogues.

Challenges and limitations you need to plan for

VTO does not solve every fit problem, and retailers who deploy it without understanding its constraints tend to be disappointed. Here is where the technology genuinely falls short.

Fit uncertainty remains

No camera can replicate the tactile experience of fabric against skin. Stretch, drape, and weight are difficult to communicate visually, and structured garments (blazers, boned corsets, stiff denim) are harder to simulate accurately than jersey or knitwear. Footwear is particularly challenging: width, arch support, and sole flexibility cannot be shown in an image.

Device and lighting constraints

Consumer-grade phone cameras vary widely in resolution and colour accuracy. Poor lighting, cluttered backgrounds, and low-resolution uploads degrade landmark detection and produce inaccurate overlays. This is a UX problem as much as a technical one: shoppers need clear guidance on how to take a usable photo.

Model bias and representativeness

Many VTO systems were trained predominantly on images of a narrow range of body types. This creates accuracy gaps for shoppers with larger bodies, non-standard proportions, or darker skin tones, where landmark detection and garment mapping can be less reliable. Retailers should test their chosen solution across a representative range of body types before launch.

Content cost

Generating 3D assets or high-fidelity product images for every SKU is expensive and time-consuming. A mid-sized fashion brand with 500 active SKUs faces a substantial content production investment before VTO can go live across the catalogue.

Data privacy and UK GDPR

This is the area where UK retailers most often underestimate their obligations. A shopper’s photo, body measurements, and biometric data are personal data under the UK General Data Protection Regulation (UK GDPR). If your VTO system processes biometric data to uniquely identify individuals, that data may qualify as a special category under Article 9, requiring explicit consent and a Data Protection Impact Assessment (DPIA).

Retailers must confirm with their VTO vendor exactly what data is collected, where it is stored, how long it is retained, and whether it is shared with third parties. A vendor that cannot answer those questions clearly is a compliance risk, not just a technical one.

Pro Tip: Before signing any VTO vendor contract, request their Data Processing Agreement (DPA) and check that data is processed within the UK or EEA, or that appropriate transfer safeguards (such as Standard Contractual Clauses) are in place. The ICO’s guidance on biometric data is the authoritative reference for UK retailers.

Quick retailer test checklist

Does the vendor provide a DPA aligned with UK GDPR?
Is biometric data processed on-device or server-side, and where are servers located?
Can shoppers opt out without losing core site functionality?
Has the system been tested on diverse body types and skin tones?
What happens to uploaded photos after the session ends?

Where virtual try-on is being used in UK e-commerce

The technology spans several distinct retail contexts, each with different integration requirements and shopper expectations.

On product pages

The most common deployment: a “Try it on” button sits alongside the standard product images. Shoppers upload a photo or activate their camera and see the item on themselves without leaving the page. This is where production try-on also fits: generating on-model imagery at scale so every shopper sees the product on a body closer to their own, even without interacting with the widget.

Virtual dressing rooms

A dedicated section of the site where shoppers can try multiple items together, build outfits, and compare looks. Higher engagement than single-product try-on, but requires more development investment.

Social commerce and video commerce

VTO assets generated from shopper photos are shareable on Instagram, TikTok, and Pinterest. Bambuser’s research highlights social sharing as a meaningful secondary benefit, extending reach beyond the original session. Live video commerce integrations allow hosts to demonstrate try-on in real time.

In-store smart mirrors

UK retailers including department stores and independent boutiques have piloted smart mirrors that let shoppers try items without undressing. The data from in-store try-ons can feed back into CRM systems, linking physical and digital behaviour.

Category-specific examples

Beauty: lipstick and eyeshadow try-on via face AR, where shade uncertainty is the primary purchase barrier. Banuba’s SDK is widely used for this.
Eyewear: 3D frame overlay via camera, allowing shoppers to see how frames sit on their face shape. Camweara and Wanna both serve this category.
Jewellery and accessories: wrist and neck AR overlays for bracelets, necklaces, and watches.
Full-body apparel: photo-based generative try-on for dresses, tops, trousers, and outerwear. Garmcheck, Swan.ai, Style.me, and OnYou all operate here.
Footwear: AR shoe try-on using foot detection; technically demanding and still maturing.
Tailoring and structured garments: 3D avatar simulation for suits and formal wear, where fit precision matters most. Perfitly and Fitnonce address this segment.

For Shopify merchants specifically, solutions like Garmcheck offer direct app integration without custom engineering, making VTO accessible to brands that lack a dedicated development team.

How to implement virtual try-on: from pilot to rollout

A structured approach prevents the most common failure mode: launching VTO on a partial catalogue with no baseline data and no clear success metric.

Step 1: Choose your integration model

Shopify app: fastest to deploy; limited customisation; best for merchants without engineering resource. Garmcheck is available as a Shopify app with a JavaScript snippet fallback.
SDK integration: more control over UX; requires front-end development; suits brands with in-house tech teams.
Hosted widget: vendor-managed embed; low maintenance; customisation varies by vendor.
Full custom build: maximum control; highest cost and longest timeline; only justified for enterprise brands with unique requirements.

Step 2: Prepare your product data

VTO quality is only as good as the product data behind it. You need:

High-resolution product images (flat lay and/or on-model, consistent background)
Accurate garment measurements per size (chest, waist, hip, length, inseam)
Fabric composition and weight (for physics simulation)
Consistent colour profiles across images

Step 3: Define your KPIs before launch

KPI What it measures Try-on rate % of product page visitors who activate VTO Add-to-cart lift Conversion rate for try-on users vs. non-users Return rate change % change in returns for VTO-assisted purchases Time on page Engagement signal; also relevant for SEO and user experience Net Promoter Score Shopper satisfaction with the try-on experience

Step 4: Run a controlled pilot

Start with one product category (typically your highest-return SKUs) and one traffic segment. Run for four to six weeks before drawing conclusions. A/B test VTO-enabled pages against control pages to isolate the effect.

Step 5: Scale catalogue coverage

Once the pilot validates your KPIs, prioritise SKU coverage by return rate. Products with the highest return volumes due to fit or sizing uncertainty deliver the fastest ROI on content production investment.

Typical timeline

Phase Duration Owner Vendor selection and DPA review 2–4 weeks Merchant + legal Technical integration 1–3 weeks Dev team or vendor Product data preparation 2–6 weeks Merchandising Pilot (one category) 4–6 weeks Merchant + analytics Catalogue rollout 8 weeks Merchandising + dev

Cost drivers to budget for

The full cost picture for VTO includes three buckets: platform licence (monthly SaaS fee, typically volume-based), asset creation (photography, 3D modelling, or AI-generated product images), and integration (one-off development or ongoing maintenance). Asset creation is usually the largest variable cost for brands with deep catalogues.

Why the competitive advantage has shifted away from the technology itself

The infrastructure for virtual try-on, including 3D body models, AI garment simulation, and real-time rendering, is becoming widely available. As Matthew Drinkwater of the London College of Fashion’s Innovation Agency argued in Vogue , the competitive advantage is shifting away from proprietary core technology and towards the quality of digital assets, the strength of brand partnerships, and the emotional resonance of the experience itself.

This has a direct implication for how UK retailers should allocate budget. Spending heavily on bespoke VTO engineering is harder to justify when capable platforms are available off the shelf. The better investment is in high-quality digital garment assets, consistent product data, and a try-on experience that reflects the brand’s identity rather than a generic widget.

The augmented reality market continues to grow across industries, and fashion is one of the most active adoption sectors. Brands that build strong digital asset libraries now will be better positioned as VTO becomes a standard feature of product pages rather than a differentiator.

Priority actions for retailers right now

Audit your product data quality: measurements, imagery, and fabric data are the foundation everything else builds on.
Invest in digital garment assets before investing in bespoke technology.
Choose a platform that integrates with your CRM (Klaviyo, for example) so try-on data feeds your retention and personalisation stack.
Link in-store and online try-on data to build unified customer profiles.
Treat the try-on experience as a brand touchpoint, not a utility feature.

Inditex’s €1.8 billion technology investment illustrates the scale at which enterprise brands are approaching this problem. Mid-market brands cannot match that spend, but they can make smarter decisions about where their digital asset investment goes.

Where virtual try-on is heading next

The technology is moving quickly, and several trends will reshape what retailers can offer within the next two to three years.

Better photorealism from generative models: diffusion-based pipelines are improving rapidly; the gap between a synthesised try-on image and a real photograph is narrowing with each model generation.
Avatar commerce: persistent digital avatars that shoppers own and use across multiple retailers, carrying their measurements and style preferences, are moving from concept to early commercial deployment.
Search integration: Google’s Shopping and visual search tools are beginning to incorporate try-on functionality, meaning VTO assets may directly influence product discoverability.
Social commerce integration: TikTok Shop and Instagram Shopping are building native try-on features; brands with strong digital asset libraries will activate these channels faster.
Lower content costs: AI-driven product photography tools are reducing the cost of generating on-model imagery at scale, removing the main barrier to full-catalogue VTO coverage.
Haptic and sensory augmentation: still experimental, but research into haptic feedback for fabric simulation is progressing; not a near-term commercial reality for most retailers.

The one-sentence recommendation: prioritise building a high-quality digital asset library and integrating try-on data with your CRM now, because those investments compound regardless of which rendering technology dominates in three years.

Making virtual try-on accessible to every shopper

Accessibility is one of the least-discussed aspects of VTO deployment, yet it directly affects whether the technology delivers on its promise of inclusivity.

Common accessibility gaps

Screen reader incompatibility: many VTO widgets are built as canvas or WebGL elements that screen readers cannot interpret, leaving visually impaired shoppers without any alternative experience.
Keyboard navigation: try-on interfaces often rely on mouse or touch interaction, with no keyboard path for shoppers who cannot use a pointer device.
Body diversity gaps: as noted in the challenges section, systems trained on narrow datasets perform less accurately for larger body types, older shoppers, and people with physical differences.
Cognitive load: complex multi-step try-on flows can be difficult for shoppers with cognitive disabilities or low digital literacy.
Device access: high-end camera requirements exclude shoppers using older or lower-specification devices.

Practical solutions

Retailers can address most of these gaps without rebuilding their VTO system from scratch. Providing a clear text alternative (size guide, fit notes, detailed measurements) alongside the try-on widget ensures that shoppers who cannot or choose not to use VTO still have the information they need. ARIA labels and keyboard navigation paths should be part of the vendor brief, not an afterthought. Testing with a diverse panel of shoppers, including those with disabilities, before launch catches the majority of UX failures. Choosing a vendor that actively maintains diverse training datasets reduces the body-type accuracy gap over time.

The UK Equality Act 2010 requires that digital services make reasonable adjustments for disabled users. A VTO widget that is the only way to access fit information on a product page is likely to fall short of that standard.

A practical perspective on where to focus your effort

The most common mistake retailers make with virtual try-on is treating it as a technology project rather than a customer experience project. The question is not “which vendor has the best rendering engine?” It is “what information does my shopper need to buy with confidence, and how do I deliver it in the fewest steps?”

For merchants ready to act:

Start with your highest-return SKUs. VTO delivers the fastest measurable ROI where fit uncertainty is already costing you money.
Get your product data right first. A well-integrated VTO system on top of poor measurement data will produce inaccurate results and erode shopper trust faster than no VTO at all.
Run a four-to-six-week A/B pilot before committing to full catalogue rollout. The data from that pilot is more valuable than any vendor case study.
Check your GDPR position before going live. The ICO’s guidance on biometric data applies to most photo-based VTO systems.
Treat the try-on output as a content asset. Images generated from shopper photos or production pipelines can support AI-driven ecommerce strategies beyond the product page itself.

For shoppers using VTO:

Upload a photo in form-fitting clothing against a plain background for the most accurate result.
Use the size recommendation alongside the visual try-on, not instead of it.
Check the retailer’s returns policy regardless: VTO improves confidence but does not eliminate all fit uncertainty.

Try virtual try-on with Garmcheck

Garmcheck is built specifically for Shopify merchants who want enterprise-grade VTO without the engineering overhead. Upload a front-facing photo, get a photorealistic garment render in under ten seconds, and receive size recommendations from eight body measurements. It integrates with Klaviyo for CRM, tracks returns and conversion analytics, and installs as a Shopify app or JavaScript snippet.

If you are evaluating VTO for your store, see how Garmcheck works or explore the fit confidence demo to see the size-recommendation features in action.

Sources

The sources below underpin the claims and figures in this article. For technical depth, the WearView glossary and Shopify’s enterprise guide are the most detailed starting points. For market context and strategic framing, the Vogue piece on VTO 2.0 and the Bambuser industry overview are the most useful.

Virtual try-on 2.0: Will it change the way we shop? | Vogue
Virtual fitting rooms: a retailer’s guide for 2026 - Shopify
Virtual Try-On: Definition & How It Works - WearView
Try before you buy: How virtual try-on is reshaping online shopping - Bambuser

Virtual try-on explained: what retailers need to know