Adding virtual try-on to Shopify: a practical setup walkthrough

Installing a try-on app on Shopify is genuinely a short job. The four steps on our Shopify integration page — install, sync, place, go live — take a morning at most. This walkthrough is about everything either side of those four steps, because that is where the difference between a working feature and a dormant one is decided.

Prepare the catalogue before you install anything

Render quality is mostly input quality. An hour spent on product data before the sync saves a week of arguing about output afterwards. Four things matter, in this order.

• Image quality. The primary image should show the whole garment, front-facing, uncropped, on a clean background, at a reasonable resolution. A lifestyle shot with the garment half out of frame is the most common cause of a poor render.
• Product type accuracy. Shopify’s product type and tags are how the system knows a garment is a skirt rather than a dress. Catalogues that have used product type as a merchandising field rather than a taxonomy field will need a pass here.
• Variants and colourways. Each colourway needs its own image attached to its variant. If three colours share one photograph, two of them will render wrong and nobody will know which.
• Size data. Per-garment measurements beat a generic brand chart by a wide margin. If you hold them in a metafield, map it during setup — this is what sharpens the size recommendation.

Then pick a starting set rather than the whole catalogue. The useful selection is the overlap of two lists: your best sellers, and your highest-return SKUs. The first gives you traffic to learn from; the second gives you the problem worth solving.

Step 1: install the app

Installation is the standard Shopify OAuth flow. The app requests read access to products and, if you want conversion reporting, read access to orders. It does not need write access to your theme files, and you should be sceptical of any app that asks for it.

Do this on a development store or with your theme duplicated if you have a large, heavily-customised setup. Not because the install is risky, but because the placement step later is much more relaxed when you can preview freely.

Step 2: sync the catalogue

The sync reads your products, images, variants and any mapped size metafields, then prepares the enabled SKUs for rendering. New and edited products come through on subsequent syncs without another manual pass.

The output worth reading is the exception list: products the sync could not categorise, variants with no image, images the system flagged as unusable. That list is your data-quality backlog, and clearing it is the highest-value work available at this point.

Step 3: choose the placement

Placement is a theme-editor job: you drop the try-on block into your product template. Where you put it changes engagement more than anything else in this list.

• Next to the size selector, above the fold, is the strongest position — it is where the hesitation already lives.
• Inside the image gallery reads as another photo and gets browsed past.
• Below the fold, under the description, performs like a footer link.
• The label matters. “See it on you” outperforms “Virtual try-on” with shoppers who have never used one, because it describes the outcome rather than the technology.

Match the block to your theme while you are there — type, radius and button style — so it reads as part of the shop rather than a bolted-on widget. A try-on panel that looks like an ad gets treated like one.

Step 4: the QA pass before you announce it

Run this list on a real device before the feature is promoted anywhere. It takes about forty minutes and it catches nearly everything.

1. 01Open five products across different categories and check the render against the product photograph for colour, print scale and hardware.
2. 02Test the hardest garments you sell: anything sheer, anything with text or a logo, anything with heavy drape.
3. 03Check a product with several colourways and confirm each variant renders its own colour.
4. 04Do the whole flow on a phone, on mobile data, and time it. Most shoppers will meet this feature on a 5-inch screen.
5. 05Test with the keyboard only: can you open the panel, change a piece and close it without a mouse?
6. 06Confirm the panel opens over your own product page and the cart survives — a try-on that redirects loses the session.
7. 07Check what happens on a product that failed the sync. The graceful outcome is that the entry point simply is not shown.
8. 08Read the shopper-facing privacy copy as a shopper would, and make sure it matches what your own policy says about uploaded photos.

Measuring impact, and when to trust it

Three metrics move, and they move at different speeds. Reading them in the wrong order is the most common way teams talk themselves out of a feature that was working.

Two rules make the numbers trustworthy. Compare cohorts of orders rather than calendar periods, so a sale week does not do the work for you. And keep a comparable set of products switched off as a control for the first quarter — the temptation to enable everything immediately is strong and it costs you the ability to prove anything.

Segment by category as well. A store-wide average will bury a strong result in denim under a flat result in t-shirts, and the denim result is the one that should decide your next rollout.

What good looks like after a month

A reasonable first-month position: try-on is on a defined product set, the exception list is cleared, engagement is measurable and stable, conversion is being compared against a control set, and nobody is drawing conclusions about return rate yet.

If you want to size the opportunity before you start, the ROI calculator takes your own order volume, return rate and processing cost and shows the arithmetic it used. It is a model, not a forecast, but it is a model you can argue with.