Adding an AI-driven personalized stylist feature to ASOS e-commerce

Global fast fashion retailer ASOS faces an average 50% return rate for online shopping, with many returned items never resold, and caused pre-tax loss of $316 million in the 2022 fiscal year. To boost its brand loyalty and reduce return rate, I designed a personalized AI stylist feature with customized avatars and tailored outfit suggestions based on individual skin tones, body types, fashion styles, and occasions to enhance customer satisfaction.

Methods
Qual interviews
Moderated usability testing

Duration
Jun 2024 (3 weeks)

Tools
Figma/FigJam, Procreate, Google form

My Roles
UX researcher
UX/UI designer

Without in-person fittings, online shoppers face significant challenges to gauge clothes fitting for them-body shape, skin color, styles, & occasions. These issues often lead to high return rates and missed opportunities for deeper customer engagement.

Problem

The ASOS AI Stylist feature, built into the e-commerce platform, addresses common online shopping pain points by combining personalization and artificial intelligent driven recommendations through three main components:

  • Personalized avatar creation for AR fitting

  • AI stylist built with a fashion style quiz to analyze user preferences

  • Suggested outfit display with expandable item details, item swapping, and cart functionality

Solution

The fast fashion industry faces high return rates

Research

I started this project with a competitive analysis. I selected Zara, H&M, Uniqlo, and Forever 21—presented in this order below—because they are fast-fashion companies that share similar target customers with ASOS. My goal was to explore how they tackle the return problem and whether they integrate the latest technology into their customer experience.

    • Fast fashion leader known for rapid trend adaptation and sleek, minimalist designs.

    • Approximately 30–50%, influenced by "bracketing" behavior where customers order multiple sizes or styles with the intention of returning some items.

    • Charges a return fee of £1.95 in the UK and $4.95 in the US for online returns via drop-off points; in-store returns remain free.

    • Limited public information on AI stylist features; primarily relies on traditional online shopping interfaces.

    • Zara uses an AR app that displays virtual models wearing new collections when users scan store displays or packages.

    • Focuses on functional, high-quality basics with an emphasis on innovation and timeless design.

    • Specific figures not publicly disclosed; however, the brand's strict return policies may contribute to a lower return rate.

    • Does not accept in-store returns for online purchases; customers must mail returns at their own expense, and the process is often considered cumbersome. Reddit+1Zara+1

    • Introduced "StyleHint," an app that provides outfit suggestions based on user-uploaded photos and preferences.

    • Targets young, budget-conscious consumers with trendy, fast-fashion offerings.

    • Specific figures not publicly available; however, the brand's broad product range and pricing strategy suggest a moderate return rate.

    • No known AI stylist features implemented; relies on traditional e-commerce browsing experiences.

    • Offers affordable fashion with a growing emphasis on sustainability and conscious collections.

    • Estimated between 20–30%, consistent with industry averages for online apparel retailers. Business of Fashion

    • Limited information available; no prominent AI stylist features reported.

    • H&M has explored AR through pop-up books and holographic experiences, notably in collaborations and its Monki brand.

I identity the following key finds from the competitive analysis:

• These brands have average 20%-50% returning rate (theguardian.com).

• Competitors had AR try-on or AI related feature but not widespread.

• None of these competitors ever implemented AI stylist feature.

I also studied existing AI stylist apps to understand user feedback and uncover inspiration to enhance ASOS’s AI stylist experience. Using a similar approach, I identified four key apps: Aiuta, Style.ai, Acloset, and Style DNA, presented in this order below.

    • Aiuta is recognized for its versatility in generating diverse outfit styles across various cultures and occasions.

    • The app offers outfit generation tailored to different cultural contexts and events.

    • The app offers outfit generation tailored to different cultural contexts and events.

    • Specific information is not publicly available.

    • Style.ai is a privately held company backed by Hanyang University Technology Holdings and the Tech Incubator Program for Startups, with seed funding of $377K.

    • The app provides personalized fashion feedback, wardrobe optimization, and accessory suggestions based on user photos.

    • $377K seed funding

    • Currently free; considering advertising

    • Acloset, developed by Looko Inc., has over 3 million users globally and is a leading AI-powered digital closet app.

    • It offers AI-driven outfit suggestions based on user schedules and weather, OOTD tracking, and digital closet management.

    • Some users report repetitive or mismatched outfit suggestions and request better logic for pairing clothes by texture and weight.

    • While exact funding figures are not publicly available, but already achieved 2.5M+ users

    • Acloset employs a freemium model with tiered subscription plans

    • Style DNA appeals to sustainability-focused users by encouraging mindful shopping and maximizing existing wardrobes.

    • It analyzes selfies to build a personal style profile and gives outfit and shopping suggestions based on fashion identity.

    • Public info on limitations is minimal; some users suggest improvements in personalization depth.

    • Specific information is not publicly available.

I identity the following key finds from the competitive analysis:

  • The onboarding process

  • These app offer outfit suggestions based on occasion, weather, and culture—consider adding event-based or seasonal styling prompts.

  • Digital closet integration is trending.

  • Some other interesting including:

Primary Research

Shoppers need outfit recommendations, diverse body & skin representation without trying on in person

To better understand the real-world challenges, habits, and unmet needs of online clothing shoppers, I conducted a survey (21 respondents) and 6 structured interviews with 2 fashion enthusiasts, 3 busy professionals, and 1 student. I learned their lifestyles, shopping behaviors, styling routines, and how they manage their wardrobes. Through this research, I aimed to learn:

How does clothing shopping fit into shoppers’ professional and personal life?

Online clothing shoppers’ shopping and styling experience/pain points?

Key Insights

Affinity map was performed to synthesize key insights. This partial view shows how I got the insights. Click the link for the full report and the image to enlarge.

I used an affinity map to synthesize key insights from the interviews. Common themes included frustration with body shape and skin tone mismatches (5/6), a strong desire for styling guidance (6/6), and a trial-and-error approach to personal styling (2/6).

Diverse body types and skin tones of models needed

Outfits recommendations for different situations needed

I identified two distinct groups, from interview and survey, and created personas to keep their traits and needs in mind throughout my design process: 1. Fashion enthusiasts value meticulous style details. 2. Busy professionals seek personal style development amidst time constraints.

Our target customers

“ How might we help online shoppers visualize outfit fit and receive personalized recommendations without being in-store? “

Building the Foundation: Onboarding Flows for Personalized Fit & Style Guidance

User Flows

To create an effective onboarding experience, the design focuses on gathering accurate fit and style data while keeping the process engaging and user-friendly. By blending adaptable conversation with familiar onboarding elements, the flow aims to minimize user fatigue and maximize personalization.

  • Designing a flow that balances conversational with traditional onboarding process to gather necessary information.

  • Implementing "skip", "pre-built database", and "save for later" options will help avoid choice overload.

  • Combining keywords and representative images is the guiding principle behind designing a visually impactful AI stylist user flow.

Avatar building user flow

AI stylist fashion quiz user flow

Click the image to see a full size image

Usability Test

Shoppers needed an even more flexible, shorter, and guided onboarding process

To understand whether the icon design effectively conveys its function, how users perceive the value of this feature within the website's content, and how they experience the feature onboarding process. To explore this, I recruited 5 users for a moderated usability test conducted on Zoom.

I aim to learn:

Feedback on AI stylist icon design

Value and experience of AI stylist onboarding process

  • AI stylist icon:

    • 5/5 users highly valued AI stylist feature.

    • Average icon search time is 5”, suggesting good visibility.

    AI stylist onboarding process

    • Average total onboarding completion is 5’ 52”, but 1-3’ is common for smooth experiences, based on NNG.

    • 4/5 users find intent driven question for body type queries confusing.

    • 5/5 users need a more flexible onboarding process, allowing skip without losing provided information

    • Click here to see the full usability test result.

Insights:

Users need shorter onboarding process

Users require more pre-built database & skippable steps

Guided onboarding is favored over intent-driven ones

Iteration

Initially, I thought key questions couldn't be skipped. However, I realized I need to be more empathetic, finding ways to reduce the user's load while still gathering the information.

1. Users need the flexibility to skip or choose from presets

BEFORE

AFTER

Break the AI stylist flow into two separate tasks: 'My Avatar' and 'My Fashion Preferences.' Users can save and edit their information and preferences, allowing them to return anytime to make adjustments for greater flexibility.

2. Break 1 long onboarding process down to 2 shorter tasks

BEFORE

AFTER

3. Change from “open-ended questions” to “guided body type questions”

During the usability test, I found that users felt more comfortable with guided questions, as they wanted the avatar to be accurately built. They felt lost with open-ended questions and were unsure how to provide accurate information without guidance.

BEFORE: ONE OPEN-ENDED QUESTION

AFTER: MULTIPLE GUIDED QUESTIONS

Personalized avatar creation for AR fitting

Basic body measurements

Basic body shape

Avatar building with tailored skin tone

Avatar body shape fine-tune

AI stylist built with fashion quiz for personalized outfit recommendation

Style keyword quiz with image support

Outfit upload for style preference building

Refining style preferences

Indicate the outfit's occasion and purpose

Suggested outfit with expandable item details and more options

Prototype

Final Thoughts

What I accomplished

  • I designed an AI Stylist feature for ASOS with personalized avatar creation for AR fitting and AI outfit suggestions.

  • This feature aims to resolve online shoppers’ challenges in matching outfits to their body types, skin tones, unique fashion styles, and occasions.

  • A seamless onboarding process balances AI intelligent adaptability and user guidance, offering flexible image options, text-image explanations, auto-saving steps, and manageable tasks while gathering sufficient information.

Next Steps and Reflection

Reflecting on my ASOS AI stylist project, I recognized the importance of guiding users through the onboarding process to ensure their comfort and trust in the platform's accuracy. Key lessons learned included avoiding standalone open-ended questions when integrating AI functionality, the effectiveness of a text-image questionnaire over text-only or image-only options, and the value of combining multiple-choice and open-ended questions. This approach offered clearer guidance for users while encouraging them to express their unique needs and preferences.

Simultaneously, I acknowledged the pressing need to delve into AI ethics within the fashion world. By promoting ethical integration and usage of AI technology, I sought to contribute positively to the industry and foster responsible innovation. Enhancing avatar accuracy and harnessing the swap function for feedback-driven outfit suggestions also became crucial components of my vision to blend technology and ethics harmoniously.

<< Previous

Effortless sustainable shopping: Responsive website + Chrome extension

Next >>

Redesigning a Background Check Website for Faster Task Completion