Designing a Personalization-First Fashion Experience for Rental Subscription Platfrom
Strategy
Product Design
Visual Design
Service Design
overview
Taelor was an early-stage B2C SaaS technology startup exploring how AI-assisted styling could reduce shopping friction and decision fatigue for male by improving early preference discovery.
The core challenge was not recommendation accuracy alone, but how to capture high-quality preference signals early enough to make personalization feel trustworthy, while operating under tight go-to-market and operational constraints.
🏁 The product operated as a fashion rental subscription service, but this case study focuses on the personalization and preference-learning architecture that powered repeat engagement over time.
Role
Founding Product Designer (0→1), owning the design vision and execution across responsive web, product, brand, and service experience, and leading research from early discovery through validation. Partnered closely with the CEO, PM, engineers, and operations to define strategy, evaluate trade-offs, and ship an MVP and service experience.
Timeline
May 2020 - Aug 2021; Multiple releases
Industry
SaaS Circular fashion
E-commerce
Outcome
The product successfully validated a personalization-first approach to fashion rental by prioritizing preference signal confidence over speed or feature breadth. Early traction demonstrated that users were more willing to trust and engage with recommendations when personalization felt intentional, guided, and progressively improved over time.
These outcomes supported customer acquisition, informed subsequent product direction, and contributed to securing pre-seed funding.
Outcome
The product successfully validated a personalization-first approach to fashion rental by prioritizing preference signal confidence over speed or feature breadth. Early traction demonstrated that users were more willing to trust and engage with recommendations when personalization felt intentional, guided, and progressively improved over time.
These outcomes supported customer acquisition, informed subsequent product direction, and contributed to securing pre-seed funding.
Approach
The core problem
Early discovery revealed that low conversion in fashion rental experiences was not driven by inventory breadth or visual design, but by low confidence in recommendation relevance and effort invested in self-exploration. Many platforms closely resembled traditional e-commerce, placing the burden of decision-making on users without providing meaningful value.
Testing different recommendation approaches showed that failures stemmed from insufficient or ambiguous preference signals captured early in the experience.
💡 This reframed the challenge from optimizing recommendation outputs to designing better input signals and confidence-building mechanisms within the system.
Approach: Outcome-driven innovation
Participants: 147 males & identified as male testers (age 26 -45)
How might we design a system where user inputs, constraints, and feedback loops work together to produce confident personalization from day one?
I approached the product as the design of a preference signal system rather than a collection of isolated features. The focus was on defining how user inputs, system constraints, and feedback loops would work together to support personalization as the product evolved.
Strategy: Optimize for Preference Signal Confidence
Building on this system framing, the design strategy prioritized preference signal confidence over speed or feature breadth.
The goal was to
increase the clarity of user intent,
reduce ambiguity in recommendations, and
design feedback loops that allowed personalization to improve through use.
This meant deliberately trading faster onboarding and early conversion for higher-quality inputs and stronger trust in recommendations—establishing a more reliable foundation for personalization as the system learned over time.
I conducted competitive analysis and interviews to setting structure of the inputs
Decision 1: Prioritize signal quality over onboarding speed
Rather than optimizing for the fastest possible onboarding typical of fashion subscription e-commerce, I designed onboarding as a guided preference discovery flow—accepting higher upfront effort to capture clearer, more discriminating preference signals.
💡 Trade-off: higher-effort onboarding → higher-quality personalization and reduced early mistrust.
Physical attributes
Quiz questions - Constrain what can work
Preference signals
Quiz questions - Express intent and frequency
Behavioral feedback
Reinforcement over time through likes, views
Reinforcement over time through feedback loops
Decision 2: Use human judgment for ambiguous preferences
Even with higher-confidence preference signals, early-stage personalization remained ambiguous—particularly when user taste was still forming and data was incomplete. The subscription model provided recurring interaction, but the core design challenge was building a reliable signal loop—capture, validation, and reinforcement—that could maintain trust while the system learned.
Rather than relying solely on automated inference, I partnered closely with PM, operations, and stylists to embed professional human judgment at moments of highest uncertainty. This allowed the experience to balance model maturity, operational cost, and user expectations while ensuring recommendations remained trustworthy during early learning phases.
I explored two approaches for professional input.
A stylist consultation early in onboarding proved most effective, helping interpret ambiguous preferences, validate early signals, and establish trust before automation scaled. This intervention improved recommendation quality while reinforcing user confidence at a critical first-touch moment.
Signal validation
(Human judgment)
Stylist consultation used to interpret edge cases and validate early preference signals. Professional judgment helped resolve ambiguity before automated recommendations scaled.
Signal reinforcement
(Operational structuring)
Operational tagging interface where stylists translated validated profile preferences into structured signals, reinforcing recommendation accuracy over time.
Learned personalization state
Learned personalization state
Learned personalization state
Personalized collections generated from accumulated signals—combining explicit input, human validation, and behavioral feedback—used to inform subsequent recommendations and shipments.
In parallel, I worked with eng to enable an online assistant widget to testing ongoing personalization support. In practice, it was primarily used for customer care, troubleshooting, and issue resolution rather than preference refinement. Based on this learning, we deprioritized the assistant as personalization and refocused human involvement where it had the highest impact.
Signal validation
💡 Trade-off: increased operational complexity → higher trust and more reliable personalization during early learning phases.
Result
The launch validated the core strategy of prioritizing preference signal quality over feature completeness.
Within four months, user engagement and satisfaction metrics indicated that deeper onboarding and human-in-the-loop personalization improved trust in recommendations, even with a limited initial inventory.
200+
First-time subscriber
/ Recorded 4 months
800+
style quiz filled /
4 months
8 / 10
NPS score
28%
reduction in onboarding bounce rate
Qualitative feedback reinforced that members felt more confident trying new styles when recommendations were guided and contextualized. External media coverage further validated the differentiated personalization approach.
“ I found myself in a slump and in need of “leveling up” my clothing game. Too often I found myself “settling” and being comfortable just wearing the “same old thing.” I was in need of some help. Taelor was just the thing. With their smart system of matching your style and mixing in new things to try, I always found myself feeling confident and rejuvenated with my outfit. Taelor matches you with a personal “stylist” that is very responsive to your needs, and always just a text away.”
Customer review
Selected media coverage:
Brand IDENTITY AND design system
I led the creation of the brand identity and foundational design system while building the product from 0→1, aligning founders, product, and marketing around a shared experience vision. I translated brand principles into a modular system of reusable components that enabled fast iteration while maintaining consistency across acquisition, onboarding, and personalization. This system became a single source of truth for design and engineering, reducing handoff friction and supporting scale as the product matured.
Reflection & Takeaways
Building a product from the ground up—and sustaining it with limited resources—was a powerful experience that stretched both creativity and execution. It was both fun and rewarding to connect digital and physical touchpoints into a cohesive customer journey. Here are a few high-level takeaways from a product design perspective:
Human-centered AI personalization: As we refined our data-driven personalization, I focused on balancing machine recommend with empathy—ensuring that even as AI evolves, human interaction remains essential to creating genuinely exclusive for personalization context.
Minimum viable product: I designed multiple concepts and features, some of which were paused and switched to built from existing tools due to technical and time constraints—an expected part of a startup environment. Staying flexible yet focused on core customer value helped the team prioritize effectively and build momentum step by step.
Inclusiveness : While the initial focus was on menswear, we built personas that also represented anyone interested in wearing men’s clothing. This inclusive mindset shaped our language, visuals, and UX writing across the app and marketing, ensuring the experience felt welcoming and relevant to all users.
Accelerating Interactive Idea Exploration [Edit 2025]: Revisiting this project during my portfolio redesign reminded me how much creativity can now be unlocked through current-generation AI tools. I see exciting opportunities to push beyond past limitations, especially where technology, personalization, and emotion intersect in shaping modern fashion experiences.
More Projects
Designing a Personalization-First Fashion Experience for Rental Subscription Platfrom
Strategy
Product Design
Visual Design
Service Design
overview
Taelor was an early-stage B2C SaaS technology startup exploring how AI-assisted styling could reduce shopping friction and decision fatigue for male by improving early preference discovery.
The core challenge was not recommendation accuracy alone, but how to capture high-quality preference signals early enough to make personalization feel trustworthy, while operating under tight go-to-market and operational constraints.
🏁 The product operated as a fashion rental subscription service, but this case study focuses on the personalization and preference-learning architecture that powered repeat engagement over time.
Role
Founding Product Designer (0→1), owning the design vision and execution across responsive web, product, brand, and service experience, and leading research from early discovery through validation. Partnered closely with the CEO, PM, engineers, and operations to define strategy, evaluate trade-offs, and ship an MVP and service experience.
Timeline
May 2020 - Aug 2021; Multiple releases
Industry
SaaS Circular fashion
E-commerce
Outcome
The product successfully validated a personalization-first approach to fashion rental by prioritizing preference signal confidence over speed or feature breadth. Early traction demonstrated that users were more willing to trust and engage with recommendations when personalization felt intentional, guided, and progressively improved over time.
These outcomes supported customer acquisition, informed subsequent product direction, and contributed to securing pre-seed funding.
Outcome
The product successfully validated a personalization-first approach to fashion rental by prioritizing preference signal confidence over speed or feature breadth. Early traction demonstrated that users were more willing to trust and engage with recommendations when personalization felt intentional, guided, and progressively improved over time.
These outcomes supported customer acquisition, informed subsequent product direction, and contributed to securing pre-seed funding.
Approach
The core problem
Early discovery revealed that low conversion in fashion rental experiences was not driven by inventory breadth or visual design, but by low confidence in recommendation relevance and effort invested in self-exploration. Many platforms closely resembled traditional e-commerce, placing the burden of decision-making on users without providing meaningful value.
Testing different recommendation approaches showed that failures stemmed from insufficient or ambiguous preference signals captured early in the experience.
💡 This reframed the challenge from optimizing recommendation outputs to designing better input signals and confidence-building mechanisms within the system.
Approach: Outcome-driven innovation
Participants: 147 males & identified as male testers (age 26 -45)
How might we design a system where user inputs, constraints, and feedback loops work together to produce confident personalization from day one?
I approached the product as the design of a preference signal system rather than a collection of isolated features. The focus was on defining how user inputs, system constraints, and feedback loops would work together to support personalization as the product evolved.
Strategy: Optimize for Preference Signal Confidence
Building on this system framing, the design strategy prioritized preference signal confidence over speed or feature breadth.
The goal was to
increase the clarity of user intent,
reduce ambiguity in recommendations, and
design feedback loops that allowed personalization to improve through use.
This meant deliberately trading faster onboarding and early conversion for higher-quality inputs and stronger trust in recommendations—establishing a more reliable foundation for personalization as the system learned over time.
I conducted competitive analysis and interviews to setting structure of the inputs
Decision 1: Prioritize signal quality over onboarding speed
Rather than optimizing for the fastest possible onboarding typical of fashion subscription e-commerce, I designed onboarding as a guided preference discovery flow—accepting higher upfront effort to capture clearer, more discriminating preference signals.
💡 Trade-off: higher-effort onboarding → higher-quality personalization and reduced early mistrust.
Physical attributes
Quiz questions - Constrain what can work
Preference signals
Quiz questions - Express intent and frequency
Behavioral feedback
Reinforcement over time through likes, views
Reinforcement over time through feedback loops
Decision 2: Use human judgment for ambiguous preferences
Even with higher-confidence preference signals, early-stage personalization remained ambiguous—particularly when user taste was still forming and data was incomplete. The subscription model provided recurring interaction, but the core design challenge was building a reliable signal loop—capture, validation, and reinforcement—that could maintain trust while the system learned.
Rather than relying solely on automated inference, I partnered closely with PM, operations, and stylists to embed professional human judgment at moments of highest uncertainty. This allowed the experience to balance model maturity, operational cost, and user expectations while ensuring recommendations remained trustworthy during early learning phases.
I explored two approaches for professional input.
A stylist consultation early in onboarding proved most effective, helping interpret ambiguous preferences, validate early signals, and establish trust before automation scaled. This intervention improved recommendation quality while reinforcing user confidence at a critical first-touch moment.
Signal validation
(Human judgment)
Stylist consultation used to interpret edge cases and validate early preference signals. Professional judgment helped resolve ambiguity before automated recommendations scaled.
Signal reinforcement
(Operational structuring)
Operational tagging interface where stylists translated validated profile preferences into structured signals, reinforcing recommendation accuracy over time.
Learned personalization state
Learned personalization state
Learned personalization state
Personalized collections generated from accumulated signals—combining explicit input, human validation, and behavioral feedback—used to inform subsequent recommendations and shipments.
In parallel, I worked with eng to enable an online assistant widget to testing ongoing personalization support. In practice, it was primarily used for customer care, troubleshooting, and issue resolution rather than preference refinement. Based on this learning, we deprioritized the assistant as personalization and refocused human involvement where it had the highest impact.
Signal validation
💡 Trade-off: increased operational complexity → higher trust and more reliable personalization during early learning phases.
Result
The launch validated the core strategy of prioritizing preference signal quality over feature completeness.
Within four months, user engagement and satisfaction metrics indicated that deeper onboarding and human-in-the-loop personalization improved trust in recommendations, even with a limited initial inventory.
200+
First-time subscriber
/ Recorded 4 months
800+
style quiz filled /
4 months
8 / 10
NPS score
28%
reduction in onboarding bounce rate
Qualitative feedback reinforced that members felt more confident trying new styles when recommendations were guided and contextualized. External media coverage further validated the differentiated personalization approach.
“ I found myself in a slump and in need of “leveling up” my clothing game. Too often I found myself “settling” and being comfortable just wearing the “same old thing.” I was in need of some help. Taelor was just the thing. With their smart system of matching your style and mixing in new things to try, I always found myself feeling confident and rejuvenated with my outfit. Taelor matches you with a personal “stylist” that is very responsive to your needs, and always just a text away.”
Customer review
Selected media coverage:
Brand IDENTITY AND design system
I led the creation of the brand identity and foundational design system while building the product from 0→1, aligning founders, product, and marketing around a shared experience vision. I translated brand principles into a modular system of reusable components that enabled fast iteration while maintaining consistency across acquisition, onboarding, and personalization. This system became a single source of truth for design and engineering, reducing handoff friction and supporting scale as the product matured.
Reflection & Takeaways
Building a product from the ground up—and sustaining it with limited resources—was a powerful experience that stretched both creativity and execution. It was both fun and rewarding to connect digital and physical touchpoints into a cohesive customer journey. Here are a few high-level takeaways from a product design perspective:
Human-centered AI personalization: As we refined our data-driven personalization, I focused on balancing machine recommend with empathy—ensuring that even as AI evolves, human interaction remains essential to creating genuinely exclusive for personalization context.
Minimum viable product: I designed multiple concepts and features, some of which were paused and switched to built from existing tools due to technical and time constraints—an expected part of a startup environment. Staying flexible yet focused on core customer value helped the team prioritize effectively and build momentum step by step.
Inclusiveness : While the initial focus was on menswear, we built personas that also represented anyone interested in wearing men’s clothing. This inclusive mindset shaped our language, visuals, and UX writing across the app and marketing, ensuring the experience felt welcoming and relevant to all users.
Accelerating Interactive Idea Exploration [Edit 2025]: Revisiting this project during my portfolio redesign reminded me how much creativity can now be unlocked through current-generation AI tools. I see exciting opportunities to push beyond past limitations, especially where technology, personalization, and emotion intersect in shaping modern fashion experiences.
More Projects
Designing a Personalization-First Fashion Experience for Rental Subscription Platfrom
Strategy
Product Design
Visual Design
Service Design
overview
Taelor was an early-stage B2C SaaS technology startup exploring how AI-assisted styling could reduce shopping friction and decision fatigue for male by improving early preference discovery.
The core challenge was not recommendation accuracy alone, but how to capture high-quality preference signals early enough to make personalization feel trustworthy, while operating under tight go-to-market and operational constraints.
🏁 The product operated as a fashion rental subscription service, but this case study focuses on the personalization and preference-learning architecture that powered repeat engagement over time.
Role
Founding Product Designer (0→1), owning the design vision and execution across responsive web, product, brand, and service experience, and leading research from early discovery through validation. Partnered closely with the CEO, PM, engineers, and operations to define strategy, evaluate trade-offs, and ship an MVP and service experience.
Timeline
May 2020 - Aug 2021; Multiple releases
Industry
SaaS Circular fashion
E-commerce
Outcome
The product successfully validated a personalization-first approach to fashion rental by prioritizing preference signal confidence over speed or feature breadth. Early traction demonstrated that users were more willing to trust and engage with recommendations when personalization felt intentional, guided, and progressively improved over time.
These outcomes supported customer acquisition, informed subsequent product direction, and contributed to securing pre-seed funding.
Outcome
The product successfully validated a personalization-first approach to fashion rental by prioritizing preference signal confidence over speed or feature breadth. Early traction demonstrated that users were more willing to trust and engage with recommendations when personalization felt intentional, guided, and progressively improved over time.
These outcomes supported customer acquisition, informed subsequent product direction, and contributed to securing pre-seed funding.
Approach
The core problem
Early discovery revealed that low conversion in fashion rental experiences was not driven by inventory breadth or visual design, but by low confidence in recommendation relevance and effort invested in self-exploration. Many platforms closely resembled traditional e-commerce, placing the burden of decision-making on users without providing meaningful value.
Testing different recommendation approaches showed that failures stemmed from insufficient or ambiguous preference signals captured early in the experience.
💡 This reframed the challenge from optimizing recommendation outputs to designing better input signals and confidence-building mechanisms within the system.
Approach: Outcome-driven innovation
Participants: 147 males & identified as male testers (age 26 -45)
How might we design a system where user inputs, constraints, and feedback loops work together to produce confident personalization from day one?
I approached the product as the design of a preference signal system rather than a collection of isolated features. The focus was on defining how user inputs, system constraints, and feedback loops would work together to support personalization as the product evolved.
Strategy: Optimize for Preference Signal Confidence
Building on this system framing, the design strategy prioritized preference signal confidence over speed or feature breadth.
The goal was to
increase the clarity of user intent,
reduce ambiguity in recommendations, and
design feedback loops that allowed personalization to improve through use.
This meant deliberately trading faster onboarding and early conversion for higher-quality inputs and stronger trust in recommendations—establishing a more reliable foundation for personalization as the system learned over time.
I conducted competitive analysis and interviews to setting structure of the inputs
Decision 1: Prioritize signal quality over onboarding speed
Rather than optimizing for the fastest possible onboarding typical of fashion subscription e-commerce, I designed onboarding as a guided preference discovery flow—accepting higher upfront effort to capture clearer, more discriminating preference signals.
💡 Trade-off: higher-effort onboarding → higher-quality personalization and reduced early mistrust.
Physical attributes
Quiz questions - Constrain what can work
Preference signals
Quiz questions - Express intent and frequency
Behavioral feedback
Reinforcement over time through likes, views
Reinforcement over time through feedback loops
Decision 2: Use human judgment for ambiguous preferences
Even with higher-confidence preference signals, early-stage personalization remained ambiguous—particularly when user taste was still forming and data was incomplete. The subscription model provided recurring interaction, but the core design challenge was building a reliable signal loop—capture, validation, and reinforcement—that could maintain trust while the system learned.
Rather than relying solely on automated inference, I partnered closely with PM, operations, and stylists to embed professional human judgment at moments of highest uncertainty. This allowed the experience to balance model maturity, operational cost, and user expectations while ensuring recommendations remained trustworthy during early learning phases.
I explored two approaches for professional input.
A stylist consultation early in onboarding proved most effective, helping interpret ambiguous preferences, validate early signals, and establish trust before automation scaled. This intervention improved recommendation quality while reinforcing user confidence at a critical first-touch moment.
Signal validation
(Human judgment)
Stylist consultation used to interpret edge cases and validate early preference signals. Professional judgment helped resolve ambiguity before automated recommendations scaled.
Signal reinforcement
(Operational structuring)
Operational tagging interface where stylists translated validated profile preferences into structured signals, reinforcing recommendation accuracy over time.
Learned personalization state
Learned personalization state
Learned personalization state
Personalized collections generated from accumulated signals—combining explicit input, human validation, and behavioral feedback—used to inform subsequent recommendations and shipments.
In parallel, I worked with eng to enable an online assistant widget to testing ongoing personalization support. In practice, it was primarily used for customer care, troubleshooting, and issue resolution rather than preference refinement. Based on this learning, we deprioritized the assistant as personalization and refocused human involvement where it had the highest impact.
Signal validation
💡 Trade-off: increased operational complexity → higher trust and more reliable personalization during early learning phases.
Result
The launch validated the core strategy of prioritizing preference signal quality over feature completeness.
Within four months, user engagement and satisfaction metrics indicated that deeper onboarding and human-in-the-loop personalization improved trust in recommendations, even with a limited initial inventory.
200+
First-time subscriber
/ Recorded 4 months
800+
style quiz filled /
4 months
8 / 10
NPS score
28%
reduction in onboarding bounce rate
Qualitative feedback reinforced that members felt more confident trying new styles when recommendations were guided and contextualized. External media coverage further validated the differentiated personalization approach.
“ I found myself in a slump and in need of “leveling up” my clothing game. Too often I found myself “settling” and being comfortable just wearing the “same old thing.” I was in need of some help. Taelor was just the thing. With their smart system of matching your style and mixing in new things to try, I always found myself feeling confident and rejuvenated with my outfit. Taelor matches you with a personal “stylist” that is very responsive to your needs, and always just a text away.”
Customer review
Selected media coverage:
Brand IDENTITY AND design system
I led the creation of the brand identity and foundational design system while building the product from 0→1, aligning founders, product, and marketing around a shared experience vision. I translated brand principles into a modular system of reusable components that enabled fast iteration while maintaining consistency across acquisition, onboarding, and personalization. This system became a single source of truth for design and engineering, reducing handoff friction and supporting scale as the product matured.
Reflection & Takeaways
Building a product from the ground up—and sustaining it with limited resources—was a powerful experience that stretched both creativity and execution. It was both fun and rewarding to connect digital and physical touchpoints into a cohesive customer journey. Here are a few high-level takeaways from a product design perspective:
Human-centered AI personalization: As we refined our data-driven personalization, I focused on balancing machine recommend with empathy—ensuring that even as AI evolves, human interaction remains essential to creating genuinely exclusive for personalization context.
Minimum viable product: I designed multiple concepts and features, some of which were paused and switched to built from existing tools due to technical and time constraints—an expected part of a startup environment. Staying flexible yet focused on core customer value helped the team prioritize effectively and build momentum step by step.
Inclusiveness : While the initial focus was on menswear, we built personas that also represented anyone interested in wearing men’s clothing. This inclusive mindset shaped our language, visuals, and UX writing across the app and marketing, ensuring the experience felt welcoming and relevant to all users.
Accelerating Interactive Idea Exploration [Edit 2025]: Revisiting this project during my portfolio redesign reminded me how much creativity can now be unlocked through current-generation AI tools. I see exciting opportunities to push beyond past limitations, especially where technology, personalization, and emotion intersect in shaping modern fashion experiences.