If you’ve looked at virtual try-on for your store this year, you’ve read the same three numbers. Conversion up 35%. Returns down 30%. Sometimes 40%. Sometimes 48%.
We build a try-on product. We had every commercial reason to want those numbers to be true. So we went and checked them.
Most of them have no source. One vendor publishes two nearly identical pages — one claiming “up to 40%” return reduction, one claiming “up to 48%” — for the same content. Another publishes “1.5x conversion increase” and “+68% conversion increase” on the same page.
We’re not naming them. The point isn’t that one vendor is sloppier than another. The point is that the entire category’s evidence base looks like this, ours included until we have something better.
Here’s the full audit — what we could trace, what we couldn’t, and what a brand doing €30k–€300k a year online should actually plan for.
Does virtual try-on reduce returns?
One retailer publishes a figure with its name attached. Zalando states that “in recent Virtual Fitting Room pilots, Zalando reduced return rates by up to 40%” — and, on the same page, that “in 2025, these solutions prevented 8% of size-related returns overall.” The 40% is a pilot ceiling. The 8% is what happened at scale. Everyone quotes the first number. Nobody quotes the second.
That gap is the most useful thing in this entire article, so sit with it for a second.
Forty percent is a real, named, first-party figure from Europe’s largest fashion platform, published on their corporate site on 22 June 2026. It is also from pilots — curated products, selected customers, controlled conditions. The eight percent is the same company, the same technology, an entire year, everything included.
A pilot ceiling is not a benchmark. When a vendor quotes you 40% and doesn’t mention the 8%, you’ve learned something about the vendor.
Two more things get dropped when the Zalando number is passed around. Their Virtual Fitting Room builds a 3D avatar from body measurements — a different mechanism from generative try-on from a photo. And Zalando has been at this since a 2022 pilot and was still moving from pilots to permanent rollout in mid-2026.
As for everyone else: we could not trace a single returns figure published by a try-on vendor to an independent study. Not one.
Our position: if a returns number appears on a vendor’s own site with no named customer and no methodology, treat it as marketing. That includes ours. We don’t publish one.
What is the actual return rate on online fashion?
It depends heavily on country and category, and the single global number you’ve seen is probably wrong. The US National Retail Federation’s 2025 Retail Returns Landscape projected 19.3% of online sales returned and $849.9B in total retail returns — that’s a forecast, for all retail, not fashion. Statista data cited by Trusted Shops (June 2022–June 2023) puts clothing at 23% and footwear at 14% as the most-returned categories.
Beyond that, honestly, the public data is thinner than the confidence with which it gets quoted.
For France we could not find a published order-level fashion return rate at all. The 39% figure that circulates in French media traces back to Statista’s Global Consumer Survey — and as far as we can establish, that survey measures the share of consumers who returned at least one item in a year, not the share of orders returned. Those are very different numbers. If someone has the primary source showing otherwise, we’d like to see it and we’ll correct this page.
For the Gulf there’s no benchmark at all. Every country-level return-rate study we checked omits the region entirely. The numbers you’ll see quoted — 25–40% for UAE fashion — come from logistics vendors’ own blogs, uncited and undated. We wouldn’t build a business case on them.
Does try-on lift conversion?
The best-sourced conversion data we found comes from DressX’s AI Virtual Try-On Report 2026, covering 1.2 million shoppers. DressX reports that try-on users were “about 3× more likely to add items to cart” and converted “50% higher.” Read it with two things in mind: DressX sells virtual try-on, and the shoppers who used it selected themselves.
We’ll take both caveats in turn, because we’d rather raise them than have you find them.
It’s a vendor’s own report. By the rule we set out above, that means it gets read with the same scepticism as anyone else’s numbers — including ours. We’re citing it because 1.2 million shoppers is a genuinely large sample and because it’s the most transparent dataset in the category, not because it’s neutral.
It’s self-selection, not a controlled trial. A shopper who chooses to try something on was already more interested in it than one who scrolled past. Some of that 3× is the try-on working. Some of it is the try-on identifying people who were going to buy anyway. Nobody in this category has published a clean A/B, and until someone does, every conversion figure in this market — ours eventually included — carries this asterisk.
How many customers actually use try-on once you install it?
This is the number that decides whether try-on pays for itself, and almost nobody in the category publishes it. The most specific public figure we found comes from a merchant writing on Substack in February 2026 about his own store experiments (“The Harsh Reality of Virtual Try-On”, 11 February 2026): 15–25% of product page visitors engaging, after optimisation. It’s one store, self-reported — but it’s more transparent than anything a vendor has published.
We’ll say the part the category avoids: a try-on button on a product page is a button. It sits there. It has no reason to be pressed. The customer has to have already found that specific item, already be willing to upload a photo of herself, and already be patient enough to wait for a render.
That’s the real problem in this category. It isn’t solved by better image quality — and every app on the Shopify App Store competing on image quality is solving the wrong thing.
What should an independent brand actually plan for?
| Claim you’ll see | What we could trace | What we’d plan for |
|---|---|---|
| Returns down 30–48% | Nothing. Untraceable across every vendor publishing it. | Nothing. Don’t put it in a business case. |
| Returns down up to 40% | Zalando, first-party, June 2026 — but explicitly from pilots | A ceiling in ideal conditions |
| Returns down 8% | Zalando, same page, full-year 2025, at scale | A more honest planning figure — and note it’s size-related returns only |
| Conversion up 35% | Vendor self-reported, no methodology | Ignore |
| Try-on users ~3× add-to-cart | DressX, 1.2M shoppers — vendor-published, self-selected | Directionally useful, heavily caveated |
| 15–25% of PDP visitors engage | One merchant’s own store, self-reported publicly | The number to measure first. Everything else depends on it. |
Do the arithmetic on your own store before you sign anything. Take your monthly product page sessions, apply an adoption rate you actually believe, and price it against a per-try-on cost.
On price: we pulled the published pricing for the main apparel try-on apps on the Shopify App Store on 24 August 2026. Entry tiers cluster around $8–$20 for 50–200 renders, mid tiers around $30–$50 for 200–500, and top tiers around $99–$149 for 1,000–1,500, with per-render overage between roughly $0.08 and $0.16. It varies more than the marketing suggests — check the actual tier against your actual volume, because the credit counts differ far more than the headline prices do.
If your adoption assumption is wrong by half, so is the whole model.
So why are we still building this?
Because we think the button is the wrong product.
A shopper who lands on a dress and presses “try it on” is at the very end of her decision. She already found the dress. The hard part happened somewhere else — usually on Instagram, usually badly.
The shopper we care about arrives earlier, with a different problem: I have a wedding in October and nothing to wear. She has an occasion, a budget, a body she knows well and a catalogue she doesn’t. That’s a conversation, not a button — and try-on is what proves the answer at the end of it, not what starts it.
That’s a claim about product design. It isn’t a conversion statistic, and we’re not going to dress it up as one.
What we’ll publish instead
We won’t put a conversion multiple on this site until we have one from a named brand, over a defined window, with a methodology we can show you in full.
Our beta partners get their own numbers from their own store, and they own them. If the numbers are bad, we’d rather find out in October than in March.
When we have them, they’ll be here — with the store name, the dates, the sample size, and the caveats.
FAQ
- Does virtual try-on reduce returns?
- Zalando, the largest first-party source, reports up to a 40% reduction in recent Virtual Fitting Room pilots and 8% of size-related returns prevented overall across 2025. No independent study confirms a category-wide effect. Treat unattributed vendor figures as marketing.
- What’s a normal return rate for an online fashion brand?
- There’s less public data than you’d expect. The NRF projected 19.3% of US online sales returned in 2025. Statista data cited by Trusted Shops puts clothing at 23% and footwear at 14%. We found no published order-level fashion return rate for France, and no benchmark at all covering the Gulf.
- Is the 39% French return rate real?
- We don’t think so. That figure traces to Statista’s Global Consumer Survey, which as far as we can establish measures the share of consumers who returned at least one item in a year — not the share of orders returned. It’s regularly repeated as a return rate in French media. If you have the primary source, send it and we’ll correct this.
- How many shoppers will actually use try-on on my store?
- Assume you don’t know and measure it in the first month. The only specific public figure we found is 15–25% of product page visitors, self-reported by one merchant in February 2026. Your adoption rate, not your conversion lift, determines whether try-on pays for itself.
- What does virtual try-on cost on Shopify?
- Published pricing as of 24 August 2026 clusters around $8–$20 for entry tiers (50–200 renders), $30–$50 mid (200–500), $99–$149 top (1,000–1,500), with overage roughly $0.08–$0.16 per render. Credit allowances vary far more between apps than the headline prices do — compare on cost per render, not on monthly price.
Sources. Zalando corporate, “How Zalando uses technology to help customers find the right size,” 22 June 2026 · NRF, 2025 Retail Returns Landscape · DressX, AI Virtual Try-On Report 2026 · Statista data cited by Trusted Shops FR, June 2022–June 2023 · “The Harsh Reality of Virtual Try-On,” Substack, 11 February 2026 · Shopify App Store listings, pricing pulled 24 August 2026.
Last updated 24 August 2026.
