Does Virtual Try-On Reduce Returns?

Short answer: possibly, but the evidence is thinner than the marketing. One randomized field experiment found that a virtual fitting tool lowered returns at an online clothing retailer. Academic work on AR try-on shows it can also do the opposite when the preview oversells the product. No independent study has yet measured the effect of AI photo try-on on return rates.

Disclosure: Genlook sells AI virtual try-on, and we do not claim that it reduces returns. This article summarizes outside research only, with every source linked.

How big the returns problem is

Online returns overall: the National Retail Federation and Happy Returns estimated that 19.3% of US online sales would be returned in 2025, in their 2025 Retail Returns Landscape (a survey of 2,006 consumers and 358 e-commerce professionals).
Online apparel: Coresight Research estimated a 24.4% return rate for US online apparel orders in 2023, from a survey of 100 brand and retailer decision-makers.
Why clothes come back: in the same Coresight survey, size and fit was the top reason (53% of respondents), ahead of color (16%) and damage (10%). The report was sponsored by 3DLOOK, a body-measurement company, which is worth keeping in mind.

So most apparel returns are about fit, and a meaningful share are about how the product looks. Those are the two things try-on tools promise to help with.

The strongest evidence: a randomized field experiment

The most rigorous study is still Gallino and Moreno (2018), published in Manufacturing & Service Operations Management. They worked with a large Latin American clothing retailer and randomly gave some customers access to a virtual fitting room, built by Metail, while a control group shopped without it.

Customers with the tool:

converted more and placed larger orders,
returned fewer items,
were less likely to order several sizes of the same item to try at home (what the authors call "home try-on").

The authors also found effects on products the tool didn't cover, which suggests it reduced general uncertainty about the store rather than only about one item.

Two caveats matter. The tool was a fit-focused virtual fitting room that used the shopper's measurements, not a generative photo of the shopper wearing the product. And the retailer's 30-day return window limited how long returns could be tracked. The study supports "fit information reduces returns", which is not the same claim as "any try-on reduces returns".

The counter-evidence: when try-on raises expectations

Two 2025 studies presented at the European Marketing Academy conference show the risk on the other side.

Hayati, Chen and van Everdingen (Erasmus University) combined secondary data with two lab experiments. AR try-on reduced returns when it showed the product's size accurately. When sizing accuracy was lacking, it inflated expectations and the outcomes were mixed.
Pape and Toporowski (University of Goettingen) studied 571 participants and found that AR try-on raises expectations. When the delivered product falls short, shoppers put more of the blame on the retailer and are less likely to come back. They call it the "AR Confidence Paradox".

The lesson is the same in both: a preview that looks better than the real product moves the disappointment from the product page to the parcel.

What the literature says overall

A 2026 integrative review by Yun and Seo in the Korean Journal of Fashion & Technology pulled together the research available through April 2026. Their conclusion: evidence on returns is promising but less mature than evidence on purchase intention, and try-on is most likely to reduce returns when it improves size and fit accuracy and narrows the gap between what shoppers expect and what arrives. They recommend judging try-on on retained sales, not conversion alone.

What about "40% fewer returns"?

You will find figures like 25%, 36%, 40% or 64% fewer returns across vendor blogs. We could not trace any of them to a published study with a control group and a stated method. Most are single client case studies, surveys of intent, or numbers repeated from other vendor pages. Treat them as marketing until the source shows who was compared with whom, over what period.

Which returns try-on can and can't affect

Return reason	Can visual try-on help?	Why
"Looks different than I expected" (style, color, how it sits)	Plausibly	Seeing the product on yourself is the gap try-on is built to close, as long as the render is faithful to the product photo.
Wrong size	Only with fit information	A photo of you wearing it shows the look, not whether a medium fits. Size needs measurements or a size chart answer.
Ordering several sizes to keep one	Plausibly	Gallino and Moreno saw less of this with fit information.
Damaged, late, wrong item	No	Operations problems, not decision problems.
Changed mind, impulse buy	Unclear	Better previews could cut doubt, or encourage more impulse orders.

This is why we treat visual try-on and sizing as two separate jobs. AI photo try-on, including Genlook's, answers "how does it look on me?". It does not measure the shopper. Pairing it with a clear answer to "which size should I take?", for example from the store's own size chart, covers the main reason clothes come back.

How to measure it in your own store

If returns are your reason for adding try-on, measure them directly instead of relying on anyone's headline number:

Run a holdout. Show try-on to a random share of visitors and hide it from the rest. Comparing shoppers who chose to use try-on against those who didn't mostly measures how engaged they were.
Compare the same products. Compare return rates per product, not store-wide, so a change in what people buy doesn't hide the effect.
Wait out the return window. Count returns only once every order in the test is past your return window.
Split by return reason. A change in "didn't look as expected" returns is the signal. Size returns tell you about your sizing information, not about the visuals.
Track retained revenue. Revenue after returns is the number that matters, since a tool that lifts conversion and returns together can still be worth it, or not.

What Genlook measures

Genlook publishes usage data from stores running try-on in a quarterly State of Virtual Try-On report. The latest edition found that 64% of orders placed after a try-on include the exact product the shopper tried on. That shows try-on is used on the items people go on to buy. It is not a returns measurement, and we don't present it as one.