Spequlo
Contact Us

The Spequlo journal

Virtual Try-On for Large Catalogues: How to Roll It Out Across Thousands of SKUs

By Divine Eyo · · 6 min read

A woman in a cream T-shirt sorts a long rail of plain knitwear in a brand's stockroom, with coffee cups on a cardboard box nearby.

Brands that ask about virtual try-on fall into two groups. One has 80 styles and wants to know if it works. The other has 4,000 styles, three colourways each and a new drop every fortnight, and wants to know how you roll it out without a six-month project. This post is for the second group.

The short answer: generative virtual try-on works from the product photos you already have, so a large catalogue is not a six-month imaging project. The real work is choosing which products to switch on first, keeping coverage up as new styles land, and measuring the effect properly. Most brands with thousands of SKUs get the best result by launching on the categories that drive returns, then expanding by data rather than by alphabet.

Why do large catalogues worry about virtual try-on at all?

Because the last generation of tools earned that worry. 3D and AR try-on needed a digital twin of every garment, per style, per colourway. At 200 SKUs that is a project. At 4,000 it is a department.

Generative try-on changed the input. As we covered in What Is Virtual Try-On?, modern apparel try-on takes a garment photo and a shopper's photo and produces a new image. The garment photo is the flat or on-model shot already sitting on your product page. No scan, no 3D asset, no reshoot.

Google's rollout is the clearest proof that this scales. When it launched try-on in the U.S. in July 2025, Google said shoppers could try on "the billions of apparel items in our Shopping Graph" by uploading a full-length photo (Google, 24 July 2025). Nobody built 3D twins of billions of items. It runs on listing imagery.

What do your product photos need to look like?

Less than you fear. The garment should be clearly visible and reasonably complete. A front-on flat lay, a ghost-mannequin shot or a clean on-model image all work. The harder cases are the ones that would be hard for a human to read too: a garment half hidden under a coat, an extreme crop, a lifestyle shot where the product is 15 percent of the frame.

Busy prints, logos and small text are the weak spot of every generative try-on tool, including ours. Fine detail can smear or shift. If much of your catalogue is heavily printed, test those styles early.

A practical audit: pull 50 SKUs across your categories, run them through the tool, and sort the results into "ship it", "print drifted" and "not usable". That hour tells you more than any vendor deck.

Which products should you switch on first?

Not all of them, and not in catalogue order. Three filters do most of the work.

Start where the returns are. Returns are the cost try-on is meant to attack. NRF and Happy Returns estimate U.S. shoppers will return $849.9 billion of merchandise in 2025, 15.8 percent of sales, with online returns running at 19.3 percent (NRF, 15 October 2025). Your own returns report will show the categories and styles that carry the most of that. Dresses, fitted tops and anything bought for an occasion usually sit near the top. Those go first.

Then where the traffic is. A try-on button on a page nobody visits does nothing. Make sure your top 200 styles by sessions are covered on day one.

Then where the margin is. A returned $40 basic and a returned $240 dress cost very different amounts to process and restock. Weight the rollout toward pieces where one saved return pays for a lot of try-ons.

Apply those three filters to 4,000 styles and a few hundred SKUs usually account for most of the returns value and most of the traffic. Launch on those and expand weekly, using the first wave's data to pick the second.

Even the biggest players do this. When ASOS launched its hybrid try-on in February 2026, it started with around 10,000 products on its iOS app for select UK and U.S. customers, with a broader rollout to follow (ASOS, 17 February 2026). ASOS carries far more than 10,000 products. A slice first, then the rest.

How do you handle colourways, variants and new drops?

This is where large catalogues actually get hurt.

Colourways. Ask whether each colour variant needs its own image, and what happens when one has no photo. Some brands shoot every colour; many shoot one and swatch the rest. Know which you are before you sign.

New drops. The failure mode is coverage decay: launch at 90 percent, add 300 styles a month, and six months later a third of your catalogue has no try-on. Ask the vendor how new products are picked up. On Shopify, a tool that reads your product images directly means a new style is covered when its photos are. Our guide to adding virtual try-on to a Shopify store covers how app embeds and app blocks pick up products, and our Shopify page covers what to check before installing.

How do you measure it across a big catalogue?

The same way as on a small one, with one advantage: you have the traffic to get an answer quickly.

Do not compare returns before and after launch. Seasonality, promotions and a new drop will swamp the signal. Run a holdout: a random share of sessions never sees try-on, the rest do, and you compare conversion and return rate between the two groups. We wrote up the method, sample sizes and the usual mistakes in How to Measure Whether Virtual Try-On Works.

With a large catalogue, also split the readout by category. Try-on may move the needle on dresses and do nothing for socks. That tells you where to expand and where to stop.

What should you ask a try-on vendor before signing?

Seven questions, in the order they tend to bite:

  1. What does a product need to be eligible, and what percentage of our current catalogue qualifies today?
  2. How are new products and new colourways picked up, and how long until they are live?
  3. How are prints, logos and sheer or layered pieces handled, and can we see our own worst cases?
  4. Can we switch try-on on and off per product, collection or category?
  5. Does pricing scale with catalogue size, with try-ons rendered, or with orders?
  6. What is the shopper's photo used for, how long is it kept, and can the shopper delete it?
  7. Can we run a holdout, and will you share the raw data?

Vague on 1 or 2, and the rollout will be slower than the sales call suggested. Vague on 7, and you will never know whether it worked.

Where Spequlo fits

Spequlo is a generative try-on layer for apparel brands. The shopper uploads one full-body photo and sees the garment on herself, using the product images already on the page. It installs via the Shopify app or a script tag, so a new style is covered when its photos are. It does the visualization only; it does not predict sizes, as we explain in Virtual Try-On vs Size Recommendation.

If you are weighing it against other tools, our comparison pages lay out the public facts side by side. The quickest test is still the simplest: pick ten styles that get returned too often and try them on yourself.

Frequently asked questions

Does every SKU need a special photo for virtual try-on?

No. Generative try-on works from the flat, ghost-mannequin or on-model images most brands already have. The exceptions are garments hard to see in the photo, heavily printed pieces, and colour variants that were never photographed.

How many products should a large catalogue launch try-on on?

Start with the styles that drive the most returns value and the most traffic, often a few hundred SKUs, then expand weekly using the data from the first wave. ASOS launched its 2026 try-on on around 10,000 products before a broader rollout.

Does virtual try-on reduce returns for big catalogues?

It can, but only a holdout test on your own traffic will tell you by how much, and the effect usually varies by category. Returns remain expensive: NRF and Happy Returns estimate 19.3 percent of online sales will be returned in 2025.

What happens to new products added after launch?

It depends on the tool. Ask how new products and colourways are picked up and how long until they are live. On Shopify, a tool that reads product images directly covers a new style once its photos are published.