September 3, 2026
Virtual try-on allows shoppers to preview frames on their own face before buying, helping eyewear brands inspire confident choices, improve conversion and reduce returns.
Amrita Bhambhani
A customer scrolling through an online eyewear catalogue faces a kind of uncertainty that does not quite exist with most other products. A jumper can be returned if the colour looks different in person. A pair of shoes can be exchanged for a half-size up. Glasses are harder to judge because they sit on the face, and measurements can only tell you so much. A frame that looks sharp on a model may sit too wide on someone with a narrower face, look different against their features or simply not suit them in the way they expected. A product page shows exactly how a frame looks on someone else, leaving the customer to guess how it will look on their own face.
That uncertainty is often what leads to a return. There may be nothing wrong with the glasses themselves, but photographs and frame measurements give a customer a rough sense of how a pair will look on their face, not always an accurate one. They may choose a shape that turns out to be too wide, a style that does not suit their features or proportions that looked different on the model. By the time they can see the difference for themselves, the glasses have already been ordered and delivered.
For eyewear that can easily cost well into three figures in the UK, this uncertainty can affect the decision at both ends of the purchase. A customer who is unsure about a frame may decide not to order it at all. Another may go ahead, only to send it back once they have tried it on.
Virtual try-on gives customers some of that information earlier, while they are still deciding on a frame, and a better sense of what might suit them. The important question is whether that added confidence leads to more completed purchases and fewer returns, and how far a virtual preview can really help someone choose the right pair.
Most virtual try-on tools start by mapping the customer's face, using a camera, either live through a webcam or from a photo they upload, to find key points such as where the eyes are, where the nose bridge sits and how wide the face is. The software uses these points to work out where a frame should sit and how big it should look, and some tools also measure pupillary distance at this stage, which is useful later if the customer orders prescription lenses.
From there, the frame gets shown in one of two ways. The first is an AR overlay, where a 3D model of the glasses sits on top of the live camera feed or photo and moves with the customer's head. The second is newer and uses generative AI, blending the frame directly into a photo instead of overlaying a 3D model, so the result looks more like a real photograph. This second method only works with a still photo rather than live video. (The same 3D-rendering approach behind AR overlays also underpins 3D product configurators used across other categories, if you want a deeper look at how that technology works and what it can return on investment.)
AR overlays handle movement and changing light well, but they need a decent camera and processor to run smoothly, and on an older phone or a slow connection the frame can lag or look stiff. Generative blending avoids this problem since it only has to process one photo, though the result stays fixed to that single pose rather than moving with the customer.
Which approach suits a brand better depends on what the shopping experience needs to do. An AR overlay suits a brand where customers browse and compare several frames quickly, while generative blending suits a brand that wants one polished image the customer might save or share.
Low Cost Glasses operates entirely online, with no physical stores anywhere in its business, which means every part of the buying journey happens on the website, without a shop assistant to offer reassurance or a mirror to check a frame against.
It had already built a clean site with a wide frame selection, though its try-on tool still relied on flat 2D images, and small mismatches in size, colour or alignment were enough to put people off, showing up as drop-offs that were a serious issue for a business with no store to fall back on when someone wasn't sure.
To fix this, they replaced the 2D images with 3D virtual try-on across the entire catalogue, using models that reflected true size, colour and texture and let shoppers rotate each frame to see how it sat on their own face from multiple angles rather than a single static shot, all within the product page itself rather than a separate step in the journey.
The results improved after the change, with monthly orders increasing from around 350 to more than 500, a 35% rise, and people spending more time exploring the site. For an online-only retailer, a better try-on experience can make it easier to choose a frame and buy with more confidence.
Virtual try-on helps customers check how a frame looks on their face, whether the shape works with their features, and whether the colour and finish match what the listing showed. These are common reasons for choosing the wrong frame online, and virtual try-on can help customers understand this before placing an order.
A 2018 study by Gallino and Moreno found that giving online customers more information about fit before they bought reduced the likelihood of returns. The research was not specific to eyewear, but the same principle applies: the better customers understand what they are ordering, the less likely they are to make a choice they later regret.
But there are limits to what a virtual preview can tell them. A manufacturing defect or the wrong frame being sent are both things a customer only finds out once the parcel arrives, not something a preview can catch beforehand. How the glasses feel after a few hours of wear falls into the same gap. A frame that looks like a good fit on screen could still pinch at the nose, feel tight at the temples or keep slipping down.
There is also the question of the prescription. A frame can look right and feel comfortable, but the glasses still need to be made to the correct measurements for the wearer's eyes. Some virtual try-on tools can estimate measurements such as pupillary distance, but getting those measurements right is a different challenge from showing how a frame looks on someone's face.
Before adding virtual try-on, eyewear brands need three things in place: accurate frame assets, reliable prescription measurements and a try-on process that's easy to use.
Catalogue and 3D assets
Every frame needs an accurate digital version for the try-on tool, either a 3D model or a set of reference images matched to the frame's real dimensions. For brands with large or frequently changing ranges, this takes time, since each new frame needs its own asset before it can appear in virtual try-on, and the size and shape have to match the real frame closely, given that customers are relying on the preview to judge whether it suits them.
Prescription measurements
For prescription glasses, brands need a reliable way to collect the measurements used to make the lenses, particularly pupillary distance, which determines where the lenses sit in front of the eyes. A 2016 study funded by the College of Optometrists found that spectacles bought online were twice as likely to fail optical tolerance standards as those bought in UK optometry practices, with inaccurate self-measured pupillary distance identified as one of the main reasons.
The same problem gets worse with progressive lenses, where different parts of the lens handle near, intermediate and distance vision, so their position needs to be precise. Brands selling progressive lenses or stronger prescriptions need a dependable way to collect these measurements online.
Accessibility
The try-on process needs to work for customers beyond those with a webcam and a good connection, since some will use a live camera while others will prefer to rely on product photos and measurements instead. Keeping clear photos, frame dimensions and fit details on the product page gives customers both options, so virtual try-on becomes one more useful part of the buying process rather than the only way to choose a frame.
Once virtual try-on is live, brands need to know whether it is helping customers choose with more confidence, complete their purchase and keep the frames they order. Here are three things worth tracking:
Conversion rate: Compare how often customers who use virtual try-on go on to buy with those who do not. This helps show whether trying on a frame is making customers more likely to complete their purchase.
Cart abandonment: Track how many customers leave after adding a frame to their basket. If virtual try-on is helping customers feel more certain about their choice, fewer of them may drop out before checkout.
Returns by reason: Look beyond the overall return rate and track why frames are being sent back. Returns because of size, shape or appearance are the most relevant here. Prescription errors, discomfort and manufacturing defects have different causes and should be tracked separately.
Buying glasses online often comes down to one question: will this frame actually suit me? Virtual try-on helps customers answer that before they buy. They can see the frame on their own face, compare different options and get a better idea of the size, shape and style they are choosing.
A few things still fall to the customer to judge. The prescription still needs to be right, and comfort is something a customer only judges once the glasses are actually worn. But it can help with one of the biggest reasons people hesitate or return a frame: it did not look the way they expected.
That's what a brand needs to know once the tool is running. Are more customers buying after using virtual try-on? And are fewer frames being returned because of their size, shape or appearance?
For brands ready to find out, Fynd's GlamAR offers 3D and AR try-on for prescription glasses, sunglasses and designer frames, with live camera, photo upload and pupillary distance capture built in, and it's built to work with a brand's existing site.
It can, for a specific kind of return. If a frame gets sent back because it looked wrong or didn't suit the customer's face, a preview beforehand can catch that. It won't reduce returns caused by an incorrect prescription, discomfort after wearing the glasses for a while, or a manufacturing fault, since none of those are things a screen can show in advance.
Comparing customers who use virtual try-on with those who ignore it can be misleading because users may already be more interested in buying. A controlled test across eligible product-page visits gives a better indication of impact. Brands should measure completed purchases, kept-order conversion and returns by reason after the full return period, rather than using try-on opens or time spent with the feature as proof of success.
AR places a digital model of the frame over a live camera feed or photograph. It can follow head movement and show the frame from different angles. Generative AI creates a new still image in which the frame is blended into the photograph. This can produce a polished result, but brands need to check that the process has not altered the product or the customer’s appearance.
Not necessarily. Browser-based virtual try-on can run within an ecommerce product page after the customer allows camera access or uploads a photograph. Device and browser support will depend on the platform being used. Brands should test the experience across common phones and browsers and retain a standard product-page alternative when it is unavailable.
How Fynd unified platform gave Klub Fox one connected system for its Mumbai warehouse and its Shopify and Myntra order flow.
Smart stores are moving beyond checkout-free shopping, using connected technology to improve pricing, stock visibility and everyday retail operations.
See how virtual try-on helps Shopify accessory brands fix low conversions, cut returns and turn browsers into buyers with AR and 3D.
Fill out the form
Share your contact information to get started
Speak to an expert
A member of our sales team will get in touch with you