Virtual try-on went from a big-retailer experiment to a Shopify app category with dozens of options in about two years. Most listings promise more conversions and fewer returns. This guide is about what the technology actually does on a product page, where it falls short, and how to tell after a month whether it earned its fee.
We build fesch, one of these apps, so we have a stake. We have tried to write the guide we would want to read as a merchant, including the parts that don't flatter the category.
What virtual try-on is
There are two different technologies sold under the same name.
Photo try-on (generative). The shopper uploads or takes a photo. The app sends that photo and your product image to an image model, which returns a picture of the shopper wearing your garment. It takes a few seconds per image. This is what almost every apparel try-on app on Shopify does today, because it works from the product photos you already have. You don't need 3D models or a photoshoot.
Live AR try-on. The camera feed is tracked in real time and a 3D model or a cut-out of the product is drawn on top. This works well for rigid products that sit on a known spot: glasses, watches, jewelry, shoes, makeup. It works poorly for clothing, which folds and drapes. AR apps usually need a 3D asset or a special image per product.
If you sell clothing, you want photo try-on. It is the closest thing to a virtual fitting room that works from the product photos you already have. If you sell mostly eyewear or jewelry, an AR app may suit you better, although photo try-on handles hats and sunglasses reasonably well too.
What it can't do
These limits apply to every photo try-on app, ours included.
- It doesn't do size advice. The model sees one photo. It doesn't know the shopper's measurements or your size chart, so the image shows how the garment looks, not whether a size M fits. A shopper can love the render and still return the jacket because the sleeves are short.
- Details drift. Small logos, printed text, fine patterns and exact buttons are the hardest part. Most renders keep them; some don't. Test your hardest product, not your easiest.
- Bodies are approximated. The garment is drawn onto the person in the photo. Loose photos, baggy clothes or unusual poses give the model less to work with, and fit can look tighter or looser than it would in reality. Good apps say so in the result ("fit may vary").
- Not every photo works. Group photos, cropped heads, very small figures, children and inappropriate images should be rejected before generation. Expect a small share of shoppers to see an error and retry.
- It takes seconds, not milliseconds. Current models need roughly 5 to 20 seconds per render, depending on the app and the load. The waiting screen matters.
- It needs traffic. A try-on button on a store with 50 visitors a month will produce a handful of try-ons. It will not create demand; it helps shoppers who are already on the product page decide.
Before you install: what to check
You can answer most of these from the App Store listing, the developer's privacy policy and the demo store.
- Render quality on your products. Look at the demo store, then run your own products through the app's onboarding. Try a print, a knit and a dark garment. Look at hands, collars and hems.
- Speed and the mobile flow. Do the whole flow on your phone: tap, upload, wait, result. Most of your shoppers will see it there.
- Where photos go and how long they stay. The privacy policy should name the processing region, the retention period and whether photos are used to train models. If you sell in the EU, ask for a data processing agreement (DPA).
- Data access. Every App Store listing has a "Data access" section. A try-on app needs your products and your theme. Some also request customer data, order history or discounts, usually for email capture or attribution features. Decide whether those features are worth that access.
- What counts as a billable try-on. Ask whether failed renders, rejected photos and timeouts count against your quota. Listings rarely say.
- Branding. Free plans usually show a "Powered by" link. Check which plan removes it.
- Languages. If you sell outside English-speaking markets, check that the widget speaks your shoppers' language.
- Theme compatibility. Most try-on apps install as an app block in the theme editor, which needs an Online Store 2.0 theme (Dawn, Horizon and most themes from the last few years).
- Controls. You will want to choose which products show the button, cap how many try-ons one shopper can make per day, and cap overage spend.
How to measure it
Install, wait four to six weeks, then look at three numbers.
Try-on rate. Try-ons divided by product page views on products with the button. This tells you whether shoppers notice and use it. Button placement matters more than anything else here: directly above or below the add-to-cart button beats the bottom of the page.
Conversion of shoppers who tried on vs. those who did not. This is the number every vendor quotes, and it is easy to misread. Shoppers who take the time to upload a photo are already more interested than average, so they would convert better even if the try-on did nothing. To get a fair read, compare against a holdout: leave the button off a few comparable collections, or alternate weeks, and compare the overall conversion rate of those products.
Returns of orders with a try-on vs. without. Returns arrive weeks after the order, so give this at least two months. If your return form asks for a reason, watch "didn't look as expected" and "didn't like it" separately from "wrong size". Try-on can only move the first two. Our German article on what virtual try-on realistically does for returns goes through the numbers.
A rough sizing example: a store with 30,000 product page views a month where 2% of viewers try on will generate about 600 try-ons a month. That is the volume to price against.
Pricing models compared
All prices are from the public Shopify App Store listings, checked on 16 October 2026. They change often, so check the listing before you decide.
| Model | Who uses it | Entry price and volume | Good for |
|---|---|---|---|
Free plan + monthly plans with overage | GenLook, Icona, AI Frame, fesch | Free tiers of 10 to 30 try-ons; paid from $14.99 to $19.99 for 80 to 200 | Testing first, then predictable monthly cost |
Monthly plans with trial, no free plan | Antla | $19.99 for 100 try-ons, 7-day trial | Stores ready to commit after a short test |
Pay-as-you-go | TryPoint (alongside free and paid plans) | $0.29 per try-on | Very low or seasonal volume |
Priced by product count | Camweara (AR) | $39 a month for up to 70 products | Jewelry, eyewear and watches with AR |
For the plan-based apps, the useful comparison is cost per included try-on at the entry tier:
| App | Entry paid plan | Included try-ons | Cost per included try-on | Overage at entry tier |
|---|---|---|---|---|
fesch | $19.99 | 200 | $0.10 | $0.12 |
Icona | $14.99 | 150 | $0.10 | $0.16 |
GenLook | $19.99 | 100 | $0.20 | $0.17 |
Antla | $19.99 | 100 | $0.20 | $0.16 |
TryPoint | $19.99 | 100 | $0.20 | $0.19 |
AI Frame | $19.99 | 80 | $0.25 | $0.16 |
Price per try-on is not the whole story. Some apps bundle email capture, size charts, Klaviyo sync or attributed revenue reporting; others put those features on higher plans. We compare features, data access and hosting in Shopify virtual try-on apps compared.
Where fesch fits
fesch is a photo try-on app for apparel stores, hosted in Frankfurt. Shopper photos and renders are deleted automatically after a few days, and a DPA is published on our site. Failed or rejected renders never count toward your try-ons, and every plan has every feature. It works for jackets, knits, shirts, dresses and tees, plus hats and sunglasses.
What fesch doesn't have yet: email capture, Klaviyo sync and revenue attribution. If those matter most to you, other apps in the table above are a better fit today.
If you want to see it on your own products, install fesch from the Shopify App Store. Onboarding renders one of your products on a model before you add anything to your theme, and the setup guide walks through the rest.
Frequently asked questions
Does virtual try-on tell shoppers which size to buy?
No. A photo-based try-on shows how a garment looks on the shopper: color, cut, length and style. It does not measure the body, so it cannot say whether a size M fits. Pair it with a good size chart or a size recommender if fit is your main return reason.
How much does virtual try-on cost on Shopify?
Most apparel try-on apps charge a monthly fee that includes a number of try-ons, plus a per-try-on overage. Entry plans sit around $15 to $20 a month for 80 to 200 try-ons. One app sells pay-as-you-go at $0.29 per try-on, and AR apps for jewelry and eyewear price by number of products instead.
How do I measure whether virtual try-on works for my store?
Track three numbers: the share of product page visitors who start a try-on, the conversion rate of shoppers who tried on versus those who did not, and the return rate of orders with and without a try-on. Shoppers who try on are already more interested, so compare against a holdout, for example collections without the button, before you credit the app.
Is virtual try-on GDPR compliant?
It depends on the app. Ask where shopper photos are processed and stored, how long they are kept, whether they are used to train models, and whether the app offers a data processing agreement. Those answers should be in the app's privacy policy and DPA before you install.