Optimizing Size Filters with UX Research

Filters within the Women’s Pants category are confusing and filtered results miss products from the intended search parameters.

Backcountry

Backcountry is a premium e-commerce outdoor retailer with the mission of connecting people to their passions. At Backcountry you’ll find the highest-performing outdoor gear as well as clothing and expert level gear knowledge. Camping, trail running, mountain biking, skiing, mountaineering, backpacking and more—all under one roof.

Role + Background

One of my Research Ops responsibilities on the UX team was monitoring the feedback tab and playback sessions in our research tool Medallia. We received a couple of comments about pant sizing being an issue and a leader additionally noticed a problem with sizing on a New Arrivals page. We noticed odd mapping of some categories of sizes and wanted to dive deeper to identify the core problem with apparel/pant sizing and create a solution that works across our many product listing page types.

When a user filters for size 5 and size 7, they are shown a different number of search results but both sizes are mapped to alphabetical size “Medium” within the backend system.

What is causing this issue?

  • Women’s pants are mapped to women’s dress sizes manually by the product content team.

  • We include odd and even women’s dress sizes in our mapping.

  • Alpha sizes are not mapped uniformly. One product’s Medium could be mapped to size 5 and another product’s Medium could be mapped to a size 7.

  • Because of inconsistent mapping, filters don’t always include the full product depth we offer in that size range.

  • Large room for human error: if the product content team does not map an alpha sized product to the size scale, the filters will display alpha sizes, and when a user selects an alpha size the results only display the singular item. It does not pull in alpha sized pants that are already mapped. This often happens with New Arrival products in particular.

This was a solution originally intended to unify outlier size scales (like European sizes) and reduce the number of filters users select to see pants in their size range (user only needs to select size 8 rather than size 8 and size Medium).

Finding a user centered solution to the problem:

As a user I want to filter for all clothing available in my size so that I can discover and purchase clothing that fits me.

Research Goal

First, understand how women describe their size for different types of pants without constraints. Secondly, when given the size scale currently offered on Backcountry, are they able to select the correct size filters to view all the pants available in their size & how confident are they in their selection?

  • Unmoderated Typeform Survey - users were able to reach out on Zoom with questions.

    20 Internal Users who shop for women’s clothing.

    We chose to utilize internal users because our employees are power users and repeat shoppers with our sites! It allowed us to quickly & cheaply product insights.

    Survey structured in two parts:

    4 open form questions asking “What size do you wear in... Dresses, Jeans, Leggings, Ski Bibs”

    Two multi select questions with the current Backcountry size filters provided

    If you were shopping for leggings, what size would you select from this menu?

    If someone’s alphabetical size was Medium, their numerical size would be ....

    We also asked users how confident they were in their filters for the first multi select answer.

  • Users think of Dress sizes in terms of Even Numerical and Alphabetical sizes.

    Users primarily think of Jean sizes in terms of Even Numerical and Waist sizes.

    Users think of Ski Bib sizes in terms of Alphabetical Sizes.

    Users primarily think of Legging and Athletic pant sizes in terms of Alphabetical Sizes. (We believe this is due to Lululemon’s numerical sizing system)
    _____________________

    Really interesting results for the question: If you were shopping for leggings, what size(s) would you select from this menu?

    • 0 users selected filters that would show all available product in their size

    • Only 4 of 20 users selected odd numerical sizes, 7 of 20 selected filters only within their range, 13 of 20 selected filters outside of their range

    • 5 of those 13 selected only filters outside of their range - 2 of those users were “Extremely Confident” in their selection

      • The 3 users with the most accurate filters to their size were not “Extremely Confident” in their answer.

    Similarly, when asked to select filters to view “Medium” sized product, only 3 of 20 users selected all of the correct sizes matched to size medium on the Backcountry site - and only one of those users selected only sizes mapped to Medium. Interestingly, 13 of the 20 users only selected even numbers.

  • Backcountry Data Findings

    On women’s pants product listing pages, size is the most interacted with facet. Users who engage with facets during their session convert at a higher rate than those who do not.

    Competitive Research
    Fashion retailers like Aritzia tend to break sizing filters into sub-categories by Alpha, Numeric, and Waist.
    Direct to Consumer retailers tend to display their sizes in one list without any transformation. We believe this is because DTC retailers have complete control over their inventory naming schemes and Skus.

    Outdoor Retailers like REI tend to show sizes in one all encompassing filter with the size range with the most products associated at the top with a “click to see more” options sizes with fewer results. Dick’s utilizes a search within filters.

  • Users think of Dresses and jeans in terms of Even Numerical and Waist Sizes.

    Users think of Ski Bib, Legging, and Athletic pant sizes in terms of Alphabetical Sizes.

    Users associate odd sizes with Juniors sizing.

User insight

I was wondering what those [odd] sizes even were. I thought only junior / girls used those “odd” numbers. So there is no mental reference point, and it’s just blindly choosing something. And then I think I’m missing something and remove all the size filters.

Key Insights

  • Users do not use odd dress sizes to size any of their clothing items.

  • There is no standard understanding of how numerical sizes correlate to alphabetical sizes.

UX Recommended Next Steps:

  • Map to and Display Even Dress & Alpha sizes in filters to reduce confusion (non-technical solution)

    00 = XXS
    0, 2 = XS
    4,6 = S
    8, 10 = M
    12, 14 = L
    16 = XL
    18 = XXL
    20 = XXXL
    22 = XXXXL

  • Consider alternative ways to unify sizing (i.e. Display Numerical & Alphabetical together)

  • Determine if a similar stream of work is necessary for Men’s Pant sizing