Rachelle Btesh - Product Designer for Growth

Stitch Fix Onboarding: The Redesign Behind a 10% Conversion Lift

Awarded Modern Retail's Award for “Best Use of AI” for StyleFile

5 MINUTES READ

Launched June 2023

Stitch Fix Onboarding: The Redesign Behind a 10% Conversion Lift

5 MINUTES READ

Launched June 2023

Company

Stitch Fix

Date

Launched June 2023

Role

Lead Product Designer

You can’t spend your way to growth if your onboarding experience isn’t inspiring, reassuring, and honestly… enjoyable.

This is the story of how we redesigned, relaunched, and rigorously A/B tested Stitch Fix’s onboarding flow — including a new “playback” moment that helped users feel understood.

I led the design effort in partnership with our design team and a huge cross-functional crew across engineering, data science, product, research, and brand.

Before we get into it, what is the Stitch Fix quiz?

It’s the entry point into Stitch Fix’s personalized styling experience.

The quiz learns a user’s style, fit, and shopping preferences, then uses that information to recommend items through advanced AI models. A stylist reviews those recommendations to create a user’s first Fix — a curated clothing box tailored just for them.

You can’t spend your way to growth if your onboarding experience isn’t inspiring, reassuring, and honestly… enjoyable.

This is the story of how we redesigned, relaunched, and rigorously A/B tested Stitch Fix’s onboarding flow — including a new “playback” moment that helped users feel understood.

I led the design effort in partnership with our design team and a huge cross-functional crew across engineering, data science, product, research, and brand.

Before we get into it, what is the Stitch Fix quiz?

It’s the entry point into Stitch Fix’s personalized styling experience.

Historical Context: How a 2012 industry-defining quiz was becoming a 2023 growth bottleneck

Stitch Fix was one of the first companies to use a long quiz-based onboarding experience to onboard users. Its 2012 launch shaped the entire category. Now, long-form quizzes or onboarding can be seen throughout every industry, and the Stitch Fix quiz really established that pattern.

But by the time I joined the Growth team in 2023, the quiz had barely changed. It had ballooned to over 70 questions and often took users 10+ minutes to complete — long enough for it to feel more like a chore than a welcome. Drop-off rates were rising, and the experience no longer reflected the brand’s ambition.

Historical Context: How a 2012 industry-defining quiz was becoming a 2023 growth bottleneck

The Problem: Marketing spend was rising, but onboarding was leaking users

Even as the company invested more in acquisition, too many new users were abandoning the quiz before ordering their first box.

Our challenge was two-fold:

Refresh the experience so it felt engaging, intuitive, and modern
Introduce a new “playback” moment to reassure users the quiz was worth their time

It was a significant investment in design, engineering, and data science — but the potential upside was massive.

The Problem: Marketing spend was rising, but onboarding was leaking users

Even as the company invested more in acquisition, too many new users were abandoning the quiz before ordering their first box.

Our challenge was two-fold:

Refresh the experience so it felt engaging, intuitive, and modern
Introduce a new “playback” moment to reassure users the quiz was worth their time

It was a significant investment in design, engineering, and data science — but the potential upside was massive.

How Onboarding Really Works: A 70-Question Quiz Feeding a Complex Recommendation Engine

Every question in the quiz provides signal to Stitch Fix’s recommendation model. That meant we couldn’t simply remove questions and hope for the best. We had to understand:

Which questions were essential
Which could be simplified
Which were outdated or redundant

So we started with a detailed audit spanning heuristics, content, data dependencies, and engineering feasibility.

How Onboarding Really Works: A 70-Question Quiz Feeding a Complex Recommendation Engine

Every question in the quiz provides signal to Stitch Fix’s recommendation model. That meant we couldn’t simply remove questions and hope for the best. We had to understand:

Which questions were essential
Which could be simplified
Which were outdated or redundant

So we started with a detailed audit spanning heuristics, content, data dependencies, and engineering feasibility.

Our Strategic Approach: Start with light, human questions before digging into style details

We reshaped the flow to feel more conversational: beginning with simple, friendly questions before gradually layering in style preferences and detailed fit guidance.

This approach:

Reduced early-session cognitive load
Built trust faster
Preserved enough signal for the model to work well

Getting there required plenty of tradeoffs and spirited debate across teams — but it produced a flow that felt substantially lighter without breaking the underlying logic.

Our Strategic Approach: Start with light, human questions before digging into style details

We reshaped the flow to feel more conversational: beginning with simple, friendly questions before gradually layering in style preferences and detailed fit guidance.

This approach:

Reduced early-session cognitive load
Built trust faster
Preserved enough signal for the model to work well

Getting there required plenty of tradeoffs and spirited debate across teams — but it produced a flow that felt substantially lighter without breaking the underlying logic.

Reimagining the Interface: A new brand, a new visual language, a new first impression

Because this work coincided with a brand refresh, we rethought everything — typography, color usage, spacing, interaction patterns — to create a more cohesive, contemporary, and approachable onboarding experience.

The result felt warmer, cleaner, and more aligned with the direction Stitch Fix was heading.

Reimagining the Interface: A new brand, a new visual language, a new first impression

The result felt warmer, cleaner, and more aligned with the direction Stitch Fix was heading.

Experience Design at Scale: A templated system for every question type

To improve consistency and speed up iteration, we built a templated design system for the quiz:

Reusable layouts
Consistent patterns
Flexible components

Early technical design reviews ensured engineering could support these patterns at scale, which became critical for future optimization rounds.

Experience Design at Scale: A templated system for every question type

To improve consistency and speed up iteration, we built a templated design system for the quiz:

Reusable layouts
Consistent patterns
Flexible components

Early technical design reviews ensured engineering could support these patterns at scale, which became critical for future optimization rounds.

Introducing StyleFile: The moment users finally saw themselves in the experience

One of the most impactful additions was StyleFile — a playback screen that reflects back what we learned about the user throughout the onboarding quiz.

It included:

Dynamic, personalized titles
A bubble chart visualizing their style preferences
A clear sense that “we get you”

StyleFile directly addressed long-standing user feedback that the quiz often felt like input without payoff. Creating an immediate moment of delight and reassurance turned out to be the primary driver of our 10% conversion lift.

Introducing StyleFile: The moment users finally saw themselves in the experience

One of the most impactful additions was StyleFile — a playback screen that reflects back what we learned about the user throughout the onboarding quiz.

It included:

Dynamic, personalized titles
A bubble chart visualizing their style preferences
A clear sense that “we get you”

Design details: My role in bringing StyleFile to life

I led the design of the bubble chart, defined the motion language through concepts and prototypes, and collaborated with our Creative team to extend the color palette for this new feature. I also designed the tap-to-expand interactions and guided leadership through multiple exploration options to shape the final direction.

Design details: My role in bringing StyleFile to life

Motion was a core part of the Stylefile experience, getting the "bubble" animation right was a cross-functional effort

Post-launch user testing and surveys showed that clients loved the bubbles—many tapped in to read more without any increase in drop-off.

This was especially meaningful because one of our core design risks was avoiding a moment that felt “too complete,” where users might stop their journey prematurely. The final design struck the right balance: engaging, exploratory, and still guiding users forward.

Motion was a core part of the Stylefile experience, getting the "bubble" animation right was a cross-functional effort

Post-launch user testing and surveys showed that clients loved the bubbles—many tapped in to read more without any increase in drop-off.

Testing & Iteration: Six rounds of A/B testing to validate and tune the experience

We ran a series of A/B tests over two months, starting with a single variant and evolving through five additional versions — each testing micro-optimizations in layout, content, pacing, and sequencing.

Across every iteration, results continued to reinforce the same story:
the new onboarding flow outperformed the old one, and the conversion lift held steady.

Testing & Iteration: Six rounds of A/B testing to validate and tune the experience

Across every iteration, results continued to reinforce the same story:
the new onboarding flow outperformed the old one, and the conversion lift held steady.

Impact: The largest conversion lift in a decade

Our redesigned onboarding flow drove a sustained 10% increase in conversion — the most significant uplift Stitch Fix had seen in more than ten years.

It improved the quality of user intent, reduced early drop-off, and became a new benchmark for how first-time experiences could directly influence business performance.

And, Stylefile won an award for "Best Use of AI"

The StyleFile playback became a standout moment in the experience, resonating with clients and gaining industry recognition.

It earned Modern Retail's “Best Use of AI” award, highlighting not only the strength of the design but also how effectively we bridged advanced models with an intuitive, human-feeling interface.