---

title: "Virtual Try-On vs Size Recommendation: What's the Difference? (2026)" slug: virtual-try-on-vs-size-recommendation description: "Virtual try-on shows how a garment looks on a body. Size recommendation predicts which size to order. They are different tools, often sold as one. Here's how each works and when a brand needs both." datePublished: 2026-09-15 dateModified: 2026-09-15 author: Divine Eyo

Virtual Try-On vs Size Recommendation: What's the Difference? (2026)

Virtual try-on generates an image of a garment on a shopper's body; size recommendation predicts which size that shopper should order. They answer two different questions — does this suit me and which one do I buy — and a brand losing money on fit returns usually needs both. The category blurs them constantly, so this guide separates them: how each works, what each needs from a brand, and where the seams show.

Contents

1. What does a size recommender actually do?
2. What does virtual try-on actually do?
3. Why doesn't try-on tell you the size?
4. Why is size recommendation from height and weight limited?
5. Which vendors do which?
6. Do you need both?
7. Where Spequlo fits
8. Questions, answered

What does a size recommender actually do?

A size recommender takes information about a shopper and information about a garment, and outputs a size. The interesting differences are in which information each vendor uses.

Body-input tools ask the shopper for a few numbers. Virtusize's own description for apparel is blunt: "Gender, age, height, weight. That's all Virtusize needs," matched against what it says are 250,000+ real fit profiles. Sizebay uses height, weight, age and a fit preference, and estimates up to nine body measurements from them. Bold Metrics runs a questionnaire that it says maps 50+ measurements — "no scanning, photos, or measuring tapes."

Outcome-data tools lean on purchase and return history rather than body inputs. True Fit describes its model as built on outcomes — "outcomes are the ground truth, everything else is an input" — across what it reports as 100M+ shoppers and $616B in transactions. That approach gets stronger with scale and weaker for a brand with a short order history.

Reference-garment tools ask a shopper to name something they already own and how it fits. Virtusize uses this for footwear and bags.

Sensor tools are the outlier: MySize (MySizeID, Nasdaq: MYSZ) measures a shopper using smartphone motion sensors rather than the camera.

What every one of them needs from the brand is the same thing: the brand's own size chart, and ideally garment measurements per size. Virtusize accepts "CSV, Excel, PDF" and normalises it. Fitle says it has indexed nearly 100,000 size guides across 9,000 brands. Without the chart, the recommender has nothing to map a body onto.

The output is a size, sometimes with a confidence signal. True Fit describes the ambition as "decision-grade fit guidance" — answering "will this feel tight in the shoulders?" rather than just "order a medium." Virtusize plots measurements onto a silhouette at the shoulder, bust and hip. None of them produces a photograph.

What does virtual try-on actually do?

Generative virtual try-on takes a photo of the garment and a photo of the shopper and produces a new image of the garment on that body. The reference architecture is Google Research's TryOnDiffusion (CVPR 2023), and Google shipped it in Shopping in June 2023.

It answers the visual question — silhouette, proportion, colour against skin, where a neckline lands, whether a midi length reads well at this height. Its inputs are images, not measurements, and its output is an image, not a number. We cover the technology in more depth in What is virtual try-on?.

Why doesn't try-on tell you the size?

Because the model is generating a plausible image, not measuring a body. It has no idea whether the garment it drew is a size 8 or a size 12; it drew a garment that fits the silhouette in the photo.

Google says this in its own Merchant Center documentation for its try-on feature: the generated image "can help people find how an item might look, but it doesn't indicate fit, suggest a size, or indicate size availability from the merchant." That is the most-deployed try-on system in the world describing its own limit, and it applies to every diffusion-based tool.

Any vendor telling you their try-on image "shows the fit" is either describing a separate size-recommendation feature they also ship, or describing something their software does not do.

→ See try-on and size guidance side by side on a product page

Why is size recommendation from height and weight limited?

Two people with identical height and weight can have different shoulder widths, torso lengths and hip-to-waist ratios. Height-and-weight tools compensate with population statistics — the most likely body for those numbers — which works on average and fails at the edges, and the edges are exactly where returns come from.

The bigger problem is that the labelled size is unreliable across brands. A vendor analysis of 1,500 brand size charts (Tellar, 2025) found the bust measurement for a UK size 10 ranged from 32 to 36 inches depending on the brand — a four-inch spread on the same label — and cited Zara at 32.5 inches against Marks & Spencer at 35. Treat the specific figures as a single vendor's dataset rather than a peer-reviewed finding, but the direction is not in dispute: a "medium" is a brand-specific term, which is why every recommender on the list above insists on the brand's own chart rather than a universal one.

Which vendors do which?

Two things to watch for when you read vendor sites. First, naming drift: Sizebay markets its size recommender under the name "virtual try-on," and its own definition says so — "Virtual try-on is a size recommendation tool." Second, ownership churn: Fit Analytics was acquired by Snap in March 2021 for roughly $124 million, then reacquired by its management and has operated independently from Berlin since January 2024. If a vendor's corporate story matters to your procurement, check the date on the press release you're reading.

Do you need both?

If your return problem is "it didn't look how I expected," try-on addresses it. If it's "it didn't fit," size guidance addresses it. PowerReviews' 2023 survey of 25,452 U.S. consumers found 39% of apparel returns were about fit and 28% about appearance — both large, and neither tool fixes the other's share.

There's also a sequencing argument. A shopper who has just seen the garment on their own body is in the highest-intent moment of the session. That is the moment to hand them a size — not to send them hunting for a chart in a modal. Tools that do only one half leave the other half on the table.

Where Spequlo fits

Spequlo does both jobs, and keeps them separate on purpose. A shopper uploads one full-body photo and sees each garment rendered on their own body; alongside it, Spequlo parses the brand's own size chart to give zone-level fit guidance — shoulders, bust, waist, hips — rather than a single letter. The size guidance comes from the brand's chart, not a universal one, for the reason in the section above. It installs on Shopify as an app embed and is free on the Shopify App Store.

→ Add try-on and size guidance to your product pages

Questions, answered

Can a virtual try-on image be wrong about size? It isn't right or wrong about size — it doesn't address size. The image will show a garment that fits the silhouette in the photo. Whether the brand's size 8 actually fits that shopper is a separate question the image cannot answer.

Is a size recommender more accurate than a size chart? It is a size chart plus a model of the shopper. Its accuracy is bounded by the chart's accuracy; a wrong chart produces confident wrong recommendations. Keep the chart current.

Which one reduces returns more? Depends on why your customers return. Look at your return-reason codes. If "fit" dominates, start with sizing. If "not as pictured" or "didn't suit me" dominates, start with try-on. Most apparel brands see both.

Do size recommenders work for a brand with no order history? Body-input tools (Virtusize, Sizebay, Bold Metrics) work from day one with a size chart. Outcome-data tools (True Fit) get better with your history and lean on cross-brand data before that.

Does Google's try-on recommend a size? No. Google's own documentation says the try-on image "doesn't indicate fit, suggest a size, or indicate size availability."

What does a brand need to supply for each? Try-on: product images meeting the vendor's photo spec. Size recommendation: the brand's size chart, ideally garment measurements per size. Both: nothing from the shopper beyond a photo or a few numbers.

---

Related: What is virtual try-on? · How to measure whether it's working · Virtual try-on for dresses