01
The Problem
$849.9 billion in returns. Fit is the largest single cause.
Online return rates are structurally higher than in-store because customers cannot try before they buy. Fashion and footwear sit at the top of the return curve — and fit is consistently the leading reason given when customers send items back.
Total US Retail Returns 2025
$849.9B
Projected 2025 total retail return value. NRF 2025 Retail Returns Landscape
Zalando — Fashion Return Rate
~50%
Approximately half of all items ordered on Europe's largest fashion platform are returned — driven by free-returns policy and multi-size ordering behaviour. Zalando Corporate
Online Return Rate 2025
19.3%
Estimated share of online sales returned in 2025, across all categories. NRF 2025 Retail Returns Landscape
Returners Citing Poor Fit
65%
Share of online shoppers who returned an item specifically because it did not fit. DealNews research via Shopify, Feb 2025
Source note: The $849.9B and 19.3% figures are from the NRF 2025 Retail Returns Landscape. The 65% fit-as-reason figure is from DealNews research cited by Shopify's February 2025 returns guide. This represents the percentage of returners who cited fit — not the share of all purchases returned. The ~50% Zalando return rate reflects Europe's higher fashion return culture driven by free-returns policies.
The core mechanics: online shoppers cannot test fit before purchase, so they order multiple sizes (bracketing) and return what does not work. Each returned unit generates a logistics cost that is rarely fully recovered. Sizing technology addresses the problem at the source — before the purchase is made.
02
Evaluation Framework
Ease of implementation vs. cost — where each solution sits
Each dot is one solution plotted on two axes — how hard it is to integrate (left = complex, right = easy) and what it costs (top = expensive, bottom = cheap). Google and Snap sit bottom-right: free, no integration needed beyond a standard product listing. WANNA and 3DLOOK sit top-left: custom SDK builds, high cost. Everything in between is API-based SaaS with subscription pricing.
9 Fashion Sizing Solutions — Implementation Complexity vs. Cost
Notes: Ease of implementation reflects technical integration complexity based on published documentation. Cost positioning is approximate — only Google Shopping VTO is confirmed free (standard product listing). All other platforms use custom enterprise or subscription pricing not publicly disclosed. AI Sizing Engines (blue) tend to sit in the mid-range on both axes. Body Scanning tools (amber) skew toward complex integration due to photo-capture flows. Virtual Try-On platforms (gold) range from free/platform to highly custom.
03
Category 1
AI Sizing & Fit Recommendation Engines
These platforms sit between the retailer's product catalogue and the customer's purchase decision, using machine learning to recommend the right size based on past purchases, body data, and garment-level fit profiles. They run as plugins or APIs — invisible to the shopper, directly integrated into the product page.
AI Sizing Engine
True Fit
SaaS — Apparel & Footwear
True Fit's platform connects shopper identity data with garment-level fit attributes across a dataset of 82 million shoppers and nearly 30,000 brands. The engine predicts size, fit preference, and likelihood of return at the individual shopper level. In November 2024, True Fit became the first — and only — size and fit app to achieve Shopify Plus Certified App Partner status. Adoption on Shopify increased 509% over the prior twelve months.
82M
shoppers in dataset
30K
brands integrated
509%
Shopify growth (12 mo)
Source: Retail Times, November 2024
View True Fit platform →
Gap Inc. Partnership — March 2026
Agentic Sizing
Bold Metrics
SaaS — AI Body Twin
Bold Metrics creates a "digital twin" of each shopper's body from minimal inputs and combines it with garment-level fit data to generate personalised size recommendations. In March 2026, Gap Inc. announced it is testing Bold Metrics' new Agentic Sizing Protocol — enabling conversational AI shopping agents to ask natural-language questions and return size recommendations. The protocol is being piloted across Gap, Old Navy, Banana Republic and Athleta.
4
Gap brands in pilot
2026
Agentic launch
Source: WWD, March 24, 2026
View Bold Metrics demo →
AI Sizing + Virtual Fitting
MySize / Naiz Fit
SaaS — Multi-market
MySize operates the Naiz Fit AI sizing platform, delivering personalised size recommendations at scale. The company reported $8.5M in revenue for 2024 (23% year-on-year growth) and targets $15M for 2025. Across 18+ countries and over 42 million size recommendations, Naiz Fit's deployed metrics show a 5.7x improvement in conversion rates and a 14% reduction in return rates. Average order value rose 27% in tracked deployments.
5.7×
conversion rate lift
−14%
return rate drop
42M+
recommendations delivered
Source: MySize via PR Newswire, 2025
View Naiz Fit demo →
Sources: True Fit certification — Retail Times, Nov 2024. Bold Metrics/Gap — WWD, Mar 2026. MySize metrics — PR Newswire.
04
Category 2
AI Body Measurement & 3D Scanning
Body scanning technology removes the measurement step that made made-to-measure inaccessible at scale. Instead of a physical fitting, shoppers take two photos on a phone. The AI extracts 80+ body measurements in under a minute — enabling size recommendations, custom pattern generation, and virtual fit simulations.
Digiday Best In-Store Technology 2024
Body Scanning AI
3DLOOK
B2B Platform — Made-to-Measure & Uniform Fitting
3DLOOK's Mobile Tailor product captures over 80 body measurements from two smartphone photos in 45 seconds. The technology uses computer vision to extract circumference, length, and volume measurements with 96–97% accuracy and 95%+ measurement repeatability. It is used by brands in made-to-measure apparel, on-demand manufacturing, and uniform fitting — including workwear and body armour (Safariland case study). The company won Digiday's Best In-Store Technology award in 2024.
80+
body measurements
45s
scan time
97%
measurement accuracy
Source: 3DLOOK official site; Digiday Technology Awards 2024
Body Measurement App
MySizeID
Consumer App + B2B API
MySizeID (part of the MySize group) offers a mobile app that measures the user's body using smartphone sensors and camera. Shoppers receive a personal size profile synced across supported retailers. The B2B API allows fashion retailers to embed size recommendations directly in their checkout flow, pulling from the user's existing MySizeID profile without requiring re-measurement. Available on iOS and Android.
iOS
App Store
Android
Google Play
Source: MySizeID official site
View MySizeID app →
Sources: 3DLOOK measurements and accuracy — 3dlook.ai. MySizeID product — mysizeid.com.
05
Category 3
Virtual Try-On: AR, AI Image Generation & 3D
Virtual try-on lets shoppers see how a specific garment looks on their own body before purchase — either through augmented reality on a live camera feed, or by uploading a photo. Platform-level deployments from Google and Snap bring this capability to billions of product listings without requiring brands to build their own tool.
Launched May 2025 — Search Labs US
AI Image Generation Try-On
Google Shopping Virtual Try-On
Platform — Integrated in Google Search
Google launched Virtual Try-On in Search Labs in the US on 20 May 2025. Shoppers upload a full-length photo and see how clothing items from Google's Shopping Graph look on their own body. The underlying model understands how different fabric materials fold, stretch, and drape on different body types. The feature works across shirts, pants, skirts, and dresses from Macy's, Kohl's, Walmart, Nordstrom and others. At launch, Google described it as "the first of its kind working at this scale, allowing shoppers to try on billions of items."
Bn
items available
May '25
launch date
Source: Google Blog, May 2025
AR Shopping Lenses
Snap / Snapchat
Platform — AR Camera (250M+ users)
Snapchat's catalog-powered Shopping Lenses enable brands to create virtual try-on experiences for apparel, footwear, and accessories directly in the Snap camera. Shoppers can try on up to 20 products in a single session. A study by Alter Agents found 80% of shoppers felt more confident in their purchase after using AR try-on. Eyewa, using Shopping Lenses, achieved 8X return on ad spend. Snap has reported 250M+ Snapchatters engaged with AR shopping lenses, with 5B+ total engagements.
80%
more confident to buy
8×
ROAS (Eyewa)
250M+
users engaged
Source: Snap for Business; Alter Agents study. Note: the 250M user and 5B engagement figures were reported in early 2022; the 80% confidence and 8X ROAS are from the same period.
View Snap Shopping Lenses →
High-heel try-on launched June 2025
3D / AR Try-On
WANNA
B2B Platform — Luxury & Premium Brands
WANNA provides 3D and AR virtual try-on for footwear, bags, jewellery, watches, scarves, and clothing. Its client base is concentrated in luxury and premium fashion: Farfetch, Dolce & Gabbana, Valentino, Balenciaga, IWC, Diesel, and Tod's. Dolce & Gabbana achieved a 6x increase in conversion rate with WANNA's implementation. In June 2025, WANNA launched virtual try-on for high-heeled shoes — the first solution to handle the specific geometry of heel elevation, supporting 7cm–9cm heel heights.
6×
conversion lift (D&G)
+9%
avg conversion increase
−4%
avg return rate drop
Source: WANNA official site
AR Fashion Platform
Zero10
Consumer App + B2B — AR Try-On & Digital Fashion
Zero10 is an AR fashion platform that enables realistic real-time virtual try-on via phone camera. Shoppers can try on digital garments, record video, and share content. Zero10 operates a direct consumer app (iOS and Android) alongside brand partnerships. Retail and fashion partnerships include Coach, Ugg, and Macy's; entertainment partnerships include Disney. In-store AR mirror installations are deployed with independent designers. The app was updated in November 2024 (v2.6.1) and April 2025.
iOS
App Store
Android
Google Play
Source: Zero10 official site; App Store/Google Play listings
View Zero10 demo →
Virtual Try-On at Retail Scale
Walmart / Zeekit
In-house Platform — Walmart.com
Walmart acquired Israeli fashion tech startup Zeekit in May 2021 and rebuilt its try-on capability as "Be Your Own Model" — allowing Walmart.com shoppers to upload a personal photo and see clothing on themselves. The feature uses machine learning to simulate shadows, fabric draping, and fit on individual body types across Walmart's private-label brands, select national brands, and marketplace sellers. Available on 270,000+ apparel items at launch and expanding.
270K+
items available
2021
Zeekit acquisition
Source: Walmart Corporate, Sep 2022
View Walmart Try-On →
Sources: Google VTO — Google Blog, May 2025. Snap — Snap for Business. WANNA — wanna.fashion. Zero10 — zero10.ar. Walmart — Walmart Corporate.
06
Category 4
In-House Technology — Fashion Retailers Building Their Own
The largest fashion retailers did not wait for third-party solutions to mature — they built proprietary sizing and virtual try-on tools and deployed them directly inside their own apps and e-commerce platforms. These in-house systems give retailers full control over the customer data generated, but require significant engineering investment to build and maintain.
Launched December 2024
3D Avatar Try-On
Zara
In-App — Zara iOS & Android
Zara's virtual try-on creates a personalised 3D avatar from a single selfie and a full-body photo taken within the Zara app. The avatar replicates the shopper's proportions and moves realistically — walking, turning — while wearing selected Zara garments. The tool launched in December 2024 across five markets: Mexico, the UK, Germany, the Netherlands, and Italy. Spain was added in a subsequent expansion. No third-party platform is involved; the technology is fully built and operated by Inditex.
5
launch markets
Dec '24
launch date
Source: Marketing4Ecommerce
Developer API launched July 2, 2025
Generative AI Try-On
Amazon
API — Amazon Bedrock (Nova Canvas)
Amazon launched its Virtual Try-On API via Amazon Bedrock (the Nova Canvas model) on July 2, 2025. Developers upload a person image and a garment image; the AI generates a photorealistic composite showing the garment on the person. The API supports apparel and accessories. Available in US, Tokyo, and Ireland regions at launch. Unlike Amazon's consumer-facing try-on feature, this API opens the underlying model to third-party retailers and developers building their own shopping experiences.
Jul '25
API launch date
3
AWS regions
Source: AWS Blog, July 2, 2025
View Amazon Nova Canvas VTO →
Launched July 2023
2-Photo Body Scan
Zalando
In-App — Zalando iOS & Android
Zalando's size-recommendation feature uses a two-photo upload (front and side view) to generate body measurements and recommend the right size for specific garments. Launched in July 2023, the tool addresses a structural challenge for Zalando: size-related returns account for approximately one third of all returns. The deployed system reduced returns by 10% on items where the feature was used, according to Zalando's own figures. Available to all Zalando shoppers via the app.
−10%
return rate (when used)
1/3
of returns are size-related
Source: Zalando Corporate, July 2023
View Zalando feature details →
Launched 2019
AR Foot Scanner
Nike
In-App — Nike App (iOS & Android)
Nike Fit uses augmented reality to scan both feet in under 15 seconds via the Nike app camera. The scan captures 13 data points per foot and generates a precise size recommendation across Nike's footwear range — accounting for differences between left and right feet, which most people have. The tool also addresses the specific fit characteristics of different Nike silhouettes. Available within the Nike app for iOS and Android since 2019.
13
data points per foot
<15s
scan time
Source: DesignBoom, May 2019
View Nike Fit details →
Launched 2018
Fit Assistant
ASOS
In-App / In-Browser — ASOS.com
ASOS Fit Assistant asks shoppers three questions — height, weight, and age — alongside purchase history analysis to generate size recommendations for specific garments. The tool cross-references the shopper's profile against return data from customers with similar body measurements who bought the same item, generating a size recommendation with confidence level. Available across ASOS.com and the ASOS app since 2018.
3
input questions
2018
launch year
Source: Marketing Dive, 2018
Try ASOS Fit Assistant →
Sources: Zara — Marketing4Ecommerce. Amazon — AWS Blog Jul 2025. Zalando — Zalando Corporate Jul 2023. Nike — DesignBoom May 2019. ASOS — Marketing Dive 2018.
5 In-House Solutions — Consumer Effort vs. Technology Generation (2018–2025)
ASOS built a simple 3-question recommender in 2018. Nike added AR foot scanning in 2019. Seven years later, Amazon launched generative AI try-on requiring a single photo upload — the top-right quadrant: most advanced tech, lowest consumer effort.
Notes: Consumer effort axis reflects the number of steps and types of input required from the shopper — answering questions (low effort) vs. submitting multiple photos or performing AR camera scans (high effort). Technology generation reflects the underlying approach: statistical recommendation model (2018), AR scanning (2019–2023), 3D avatar or generative AI (2024–2025). Amazon Jul 2025 = Nova Canvas VTO API via Amazon Bedrock. Sources: official brand pages and AWS documentation referenced in the cards above.
07
Comparison
Solution Matrix — 9 Companies Compared
The nine solutions split across three technology categories, two deployment models (platform-level and brand-level), and two primary customer segments (mass market and luxury/premium). The matrix maps each company on the dimensions that matter most for a retailer evaluating fit-tech adoption.
Deployment types — what each model means
SaaS plug-in
Installed from an app marketplace (e.g. Shopify App Store) with no code. Activated with a few clicks. Vendor manages updates and infrastructure. Fastest time-to-value for retailers already on the supported platform.
SaaS / API
Cloud subscription accessed via REST or GraphQL API. Requires a developer integration — typically 1–4 weeks of engineering work. More flexible than a plug-in but demands technical resource. Vendor still manages all infrastructure.
B2B API
Direct API integration with greater flexibility and customisation than a SaaS plug-in. Requires developer resources and longer onboarding. Typically used for enterprise clients, uniform fitting, or custom manufacturing pipelines where standard plug-ins don't fit.
B2B SDK / API
A software development kit (SDK) bundled with API access. The SDK handles the 3D rendering, AR camera, or body-scan interface; the API carries the data. Higher integration complexity and time-to-deploy. Typically used by luxury brands building custom try-on experiences.
Platform (Google / Snap / Walmart)
Built into the platform. Brands access the feature automatically by listing products on Google Shopping, running Snap ads, or selling on Walmart.com — no separate technical integration required. The platform handles everything; the brand just provides product listings.
App + B2B API
Two-sided model. The consumer-facing experience runs through a mobile app that users download; retailers connect via API to pull size profiles or recommendations. Both the shopper and the retailer must adopt for it to function.
| Company | Category | Brands using | How It Works | Key Metric | Target Market | Deployment | App | Latest (2024–25) |
|---|---|---|---|---|---|---|---|---|
| True Fit | AI Sizing Engine | Macy's, Lululemon, Brooks Brothers, PacSun, Dick's Sporting Goods, Lucy & Yak | Matches shopper purchase history to 82M-shopper dataset + 30K brand profiles | 509% Shopify growth in 12 months | Mass market, mid-market apparel & footwear | SaaS plug-in | — | Shopify Plus Certified, Nov 2024 |
| Bold Metrics | AI Sizing Engine | Gap, Old Navy, Banana Republic, Athleta | Digital body twin from minimal inputs; conversational AI sizing (Agentic Protocol) | Gap Inc. pilot across 4 brands (2026) | Mid-market to premium apparel | SaaS / API | — | Agentic Sizing Protocol, Mar 2026 |
| MySize / Naiz Fit | AI Sizing Engine | Not publicly disclosed (18+ countries, 42M+ recommendations delivered) | AI size recommendation at checkout; 42M+ recommendations across 18 countries | 5.7× conversion lift; −14% returns | Multi-market, mass to mid | SaaS / API | iOS + Android | $8.5M revenue 2024; targeting $15M (2025) |
| 3DLOOK | Body Scanning | 1822 Denim, Baynes & Baker, Safariland Group, Bettter | 2-photo AI scan → 80+ body measurements in 45 seconds; 96–97% accuracy | 95%+ measurement repeatability | Made-to-measure, uniform, workwear | B2B API | — | Digiday Best In-Store Tech 2024 |
| MySizeID | Body Scanning | Retailer clients not separately disclosed (part of MySize group) | Smartphone sensor + camera measurement; synced size profile across retailers | — | Consumer-facing; mid-market retail | App + B2B API | iOS + Android | Part of MySize group |
| Google Shopping VTO | Virtual Try-On | Macy's, Kohl's, Walmart, Nordstrom (at launch, May 2025) | Upload photo → AI drapes billions of Shopping Graph items on your body | Billions of items at launch (May 2025) | All — platform-level reach | Platform (Google Search) | — | Launched May 2025 (US, Search Labs) |
| Snap AR Shopping | Virtual Try-On | Eyewa (8× ROAS confirmed); available to any brand running Snap Shopping ads | AR camera lens; up to 20 products tried in one session from brand catalogue | 80% shoppers more confident; 8× ROAS | Mass market; youth/social audience | Platform (Snapchat) | Snapchat app | AR glasses with shopping planned 2026 |
| WANNA | Virtual Try-On | Dolce & Gabbana, Valentino, Balenciaga, Farfetch, IWC, Diesel, Tod's | 3D/AR try-on for footwear, bags, jewellery, clothing; luxury brand integrations | 6× conversion (Dolce & Gabbana); +9% avg | Luxury & premium (Valentino, Balenciaga, Farfetch) | B2B SDK / API | — | High-heel try-on launched Jun 2025 |
| Zero10 | Virtual Try-On | Coach, Ugg, Macy's, Disney | Real-time AR camera try-on; digital fashion; in-store AR mirrors | Coach, Ugg, Macy's, Disney partnerships | Consumer-facing; retail & entertainment | App + B2B | iOS + Android | v2.6.1 Nov 2024; update Apr 2025 |
| Walmart / Zeekit | Virtual Try-On | Walmart only (in-house — Zeekit acquired May 2021) | Photo upload → AI simulates clothing drape on personal body; Walmart.com native | 270,000+ items available | Mass market — Walmart shoppers | Platform (Walmart.com) | — | Acquired 2021; scaled 2022–25 |
Matrix sources: True Fit — Retail Times Nov 2024; Bold Metrics — WWD Mar 2026; MySize — PR Newswire 2025; 3DLOOK — 3dlook.ai; MySizeID — mysizeid.com; Google — Google Blog May 2025; Snap — Snap for Business; WANNA — wanna.fashion; Zero10 — zero10.ar; Walmart — Walmart Corporate Sep 2022.
08
Market Signal
Three structural shifts across the nine solutions
1. Platform-level virtual try-on removes the build-your-own barrier. Google and Snap deploy try-on at platform scale — any brand listing on Google Shopping or running Snap ads can reach shoppers with try-on capability without building proprietary technology. For mid-market brands without an R&D budget for AR, this is a structural change in what is achievable.
2. AI sizing is moving from recommendation to agentic. Bold Metrics' Agentic Sizing Protocol and its Gap partnership represent the next stage: sizing recommendations delivered through conversational AI agents, not a product-page widget. As AI shopping assistants expand, the sizing engine sits inside the agent — not on the page.
3. The returns problem has not been solved — but the tools are production-ready. MySize's 14% return rate reduction and WANNA's tracked conversion lifts show that deployed tools are moving measurable outcomes. The gap is adoption: most mid-market retailers have not integrated any sizing technology. The NRF's 19.3% online return rate will persist until the adoption curve closes.