Improving TTF's Fit Prediction Widget

Improving TTF's Fit Prediction Widget

Improving TTF's Fit Prediction Widget

This is the story of how I led the charge on collecting and synthesizing qualitative data during a usability study—and how those insights helped shape key product decisions.

This is the story of how I led the charge on collecting and synthesizing qualitative data during a usability study—and how those insights helped shape key product decisions.

This is the story of how I led the charge on collecting and synthesizing qualitative data during a usability study—and how those insights helped shape key product decisions.

background

TrueToForm's 3D Fit Widget

Ever second-guessed what size to order when shopping online? You’re not alone—42% of online returns happen due to fit issues, with many shoppers resorting to bracketing (buying multiple sizes to try at home). For retailers, this translates to billions in lost revenue annually.

TrueToForm (TTF) is solving this problem with its 3D Fit Widget, a tool that predicts garment fit based on body measurements. Originally designed for made-to-measure clothing, TTF now helps shoppers confidently find their ideal size through:

Survey Avatars: AI-generated avatars based on similar body data.
Body Scans: Personalized avatars created from a quick phone scan for precise sizing.
the challenge

Turning Scattered Feedback into Product Direction

While the broader study aimed to validate the fit algorithm’s accuracy, my focus was on understanding the human experience behind the numbers. My qualitative research goals were to:

– Explore how participants felt about their recommended size
- Uncovering Pain Points & Areas for Improvement
– Gather feedback on the visual experience of the avatar
– Translate user insights into actionable product improvements

my role

Assistant UX Researcher / UX Designer

I partnered with the lead UX researcher to support a usability study from end to end. My focus was on the qualitative side:

  • Facilitating remote testing sessions

  • Collecting and organizing user feedback

  • Leading affinity mapping sessions

  • Synthesizing insights and spotting usability patterns

  • Presenting key findings and design recommendations to stakeholders

With a small team of two researchers, two engineers, and the Co-Founders, we tackled a massive initiative to improve the 3D Fit Widget—navigating tight budgets, an ambitious timeline, and plenty of surprises along the way.

the approach

Virtual Fit Prediction vs Reality 

We ran moderated usability sessions with 27 participants, comparing real-life garment fit to the algorithm’s predictions—using both survey data and 3D body scans.

While the broader goal was to assess accuracy, I focused on the qualitative side: observing participant reactions, digging into their expectations vs. reality, and capturing feedback in real time. From there, I organized insights and surfaced recurring themes around sizing confusion, language clarity, and trust in the tool.

And because no research project is without its challenges, we made it all happen on a tight budget and timeline. (How? Let’s just say creativity and resourcefulness were ✨essential✨ )

Kicking off Testing

We ended up conducting 27 moderated tests via zoom with women in the USA, between ages 18-40, who frequently shop for clothing online and have experience using size prediction tools.

Pre-test instructions

Before the session, participants completed a body scan in the app—giving us a baseline to explore how they felt about the fit predictions during testing.

Mailing test materials

To explore size perception and fit preferences, we sent each participant three sizes of the same shirt—plus a measuring tape to support hands-on feedback.

Moderated testing

During the session, we guided participants through taking their own body measurements—opening up conversations about accuracy, comfort, and how confident they felt in the tool’s recommendations.

Recording observations

As participants tested each size, I took the lead on observing their interactions and capturing qualitative feedback—listening for moments of hesitation, confusion, or delight. My notes focused on trust, usability, and how well the experience matched their expectations.

27 participants, 81 T-shirts, and six weeks of testing later—we’ve got answers, some setbacks, and plenty to reflect on.

27 participants, 81 T-shirts, and six weeks of testing later—we’ve got answers, some setbacks, and plenty to reflect on.

27 participants, 81 T-shirts, and six weeks of testing later—we’ve got answers, some setbacks, and plenty to reflect on.

The Discovery

📊 The Big Picture

While my focus was on qualitative insights, our team’s work led to meaningful improvements—boosting the fit algorithm’s accuracy from 85% to 92%. By spotting issues like incorrect avatar shoulder widths, we helped refine the tool, build shopper trust, and set the stage for fewer returns and wider adoption.

But numbers only told part of the story—here’s what I uncovered from users themselves.


Exploring How Participants Felt About Their Recommended Size

👚 Fit Preferences Weren’t One-Size-Fits-All

As I gathered user feedback during testing, one thing became obvious: “perfect fit” is deeply personal.

Some participants preferred a snug, form-fitting look, while others leaned towards a looser, more relaxed fit—making it really hard to get a success rate - proving that fit is more personal than just numbers.

👱🏻‍♀️ "I’d buy the 2X because even though it’s a little, baggy, it’s more comfortable this way. I like my shirts to be on the larger side"
👩🏽‍🦱 “I prefer my clothing to fit tighter, so I usually choose smaller sizes to avoid excess fabric.”

These insights pushed us to reconsider how success is defined and how we communicate fit.



Uncovering Pain Points & Areas for Improvement

📝 Fit language caused confusion

Users found terms like "tight" and "loose" confusing since they’re not commonly used in fashion.

So, midway through testing, we updated the language to better match industry norms:

While we didn’t have exact metrics, confusion noticeably dropped and participants responded more confidently to the updated descriptions.


Gather feedback on the visual experience of the avatar

👤 Avatars Felt Bland & Lifeless

Users describe the avatars as bland, clinical, and lacking personality, comparing it to "naked mannequins" or a "TSA scan."

👩🏼‍🦰 “Nothing excites me about this avatar. It reminds me of a medical app.”


To explore a better alternative, I used AI and Photoshop to create avatars that feel more engaging and approachable—without being uncomfortably realistic. 

These explorations aim to breathe life into the avatars because shopping should be fun, not bland and clinical!

b

In summary, identifying an algorithm error helped us dramatically boost the tool’s accuracy—but user feedback reminded us there’s still work to do to make sizing feel more personal and engaging.

Reflection

What We Learned

Looking back, we might have taken on a bit more than we could handle with just two researchers and limited resources. The first surprise? The cost of recruiting our target of 30 participants was eye-opening. We wanted to streamline the process with a user recruitment platform, but the price tag was way beyond our budget—especially since we had to provide each participant with three T-shirts and a measuring tape.

So, we rolled up our sleeves and got creative! I crafted social media ads targeting our ideal demographic, which was successful but also involved a lot of scheduling and admin work that ate into our tight timeline.

Another curveball was ensuring participants measured themselves accurately while we guided them remotely. Without control over their environment, we often questioned the accuracy of their measurements.

Identifying true patterns in discrepancies was challenging due to several factors:

  • Variability in T-shirt sizes

  • Lack of oversight in the measuring process

  • Inconsistent participant assessments

  • Researcher bias—sometimes participants insisted a loose-fitting shirt was fitted, leaving us to make judgment calls.

What We Would Do Differently

In hindsight, we’d create a more controlled environment and experiment with different materials, like stretch vs. non-stretch, to improve accuracy.

What’s Next?

As the developers focus on refining the backend, we’re excited to leverage the valuable qualitative data we gathered during testing to improve V1 of the 3D fit widget. With our research in hand, we’re prepared to make some impactful enhancements!

Need a thoughtful designer?

Get in touch

Need a thoughtful designer?

Get in touch

Need a thoughtful designer?

Get in touch
Black and white avatar

Joanna Corona

UX/UI Designer

Joanna is a NYC-based product designer, most recently at fashion tech startup TrueToForm, where adaptability and innovation drove her work.

Black and white avatar

Joanna Corona

UX/UI Designer

Joanna is a NYC-based product designer, most recently at fashion tech startup TrueToForm, where adaptability and innovation drove her work.

Black and white avatar

Joanna Corona

UX/UI Designer

Joanna is a NYC-based product designer, most recently at fashion tech startup TrueToForm, where adaptability and innovation drove her work.