Card sorting before pixels: how one study shaped what shoppers searched for
When I started redesigning The Sustainable Sequin Company's online shop, the obvious problem was visual. The real problem turned out to be vocabulary.
TSSC sells recycled and biodegradable sequins to fashion designers, costume makers, embroiderers and independent makers. The catalogue is big and unusually detailed: colours, shapes, sizes, finishes, quantities and materials all matter to the buyer. The UX audit I ran at the start of the project found a broken search, limited filtering, unclear quantities and navigation that used overlapping terms for the same things.
This post is about the step that fixed the structure underneath all of that: building a new information architecture and testing it with card sorting before designing a single screen. It was part of my MSc major project, run as a real 16-week internship, and the store is now live.
Why I didn't start with the interface
It's tempting to open Figma on day one. The old site looked dated, so a new look felt like the job.
But the research said otherwise. Across two surveys (one for the target audience, one for past customers) and a set of semi-structured interviews, people kept returning to the same needs: clear categories, working filters and search, accurate colours and a sense of scale. Nobody asked for a prettier homepage. They asked to find the right sequin quickly and be sure it was the right one.
A new skin on the old structure would have kept every one of those problems. So the structure came first.
How the old navigation got in the way
Specialist shops tend to grow their menus around how the business thinks about stock, not how customers think about what they need. TSSC's navigation had grown the same way. Several categories overlapped, and a shopper looking for, say, a small metallic round sequin had more than one plausible place to start and no guarantee that any of them held everything.
That kind of overlap is expensive. Every time a user has to guess between two labels, some of them guess wrong, decide the product isn't stocked, and leave.
Building a structure from how customers search
I drafted a new information architecture organised around the attributes customers actually use to decide: product type, material, and the specific properties that matter for this product, such as shape, size and finish.
Then I tested it with card sorting. Participants grouped product cards into categories and named the groups themselves. The point wasn't to get a perfect consensus. It was to see where people agreed without help, where they hesitated, and which words they reached for naturally.
A few patterns came through, from the card sort and the research before it:
People sort by attribute, not by collection. Shape, size and material were the natural groupings, so those became filters rather than being buried inside category names.
Precise terms beat clever ones. Makers use exact language (sizes in millimetres, hole positions). Labels needed to match that language rather than marketing names.
Sustainability is a property, not a section. Users wanted to know what each product was made of while browsing, not on a separate page.
The structure was refined from those results into the final navigation and filter set.
Testing the screens on top of it
With the structure in place, I designed the store in Figma and ran two rounds of usability testing. Round one surfaced the remaining friction in navigation, product information and interactions. I revised the design and tested again in round two to validate it and identify what was still open.
Then I built it. I implemented the redesign in Shopify myself, including the product structure that a catalogue like this needs: many variants across colours, shapes, sizes, finishes, quantities and materials, plus inventory, shipping, international markets and SEO.
The validation I didn't plan for
The most satisfying evidence came after launch, from real behaviour rather than a study.
In the first month live (21 September to 20 October 2026), the top on-site searches were terms like metallic, circles, centre hole, 5mm and 8mm. Those are exactly the attributes the card sort had pointed to, and exactly the filters the new structure exposes. Customers were searching in the shape of the information architecture.
That month the store took 29 orders in 30 days: 11 on the new website and 18 through the connected Etsy shop. The website had 1,007 sessions, with 36 add-to-baskets, 16 shoppers reaching checkout and 9 completed checkouts. Mobile visitors added to basket nearly twice as often as desktop visitors (4.9% vs 2.6%), and visitors from Google converted at 1.5%, with all three Google sales coming through Merchant Center listings.
One month is a small sample, and I'm treating these numbers as a baseline rather than a verdict.
What I'd do next
The funnel makes the next problem clear. The biggest drop is between browsing and adding to basket: of 1,007 sessions, only 3.6% added something. Once people commit, they move on well (44% of basket adds reached checkout, and 56% of those completed it).
So the next test is product comparison. In the second round of usability testing, comparing similar products was the issue still open. For a catalogue where two items differ by a millimetre or a finish, letting people see options side by side is likely the biggest remaining lever. I'd prototype a lightweight comparison view, test it with a handful of makers, and measure the change in add-to-basket rate.
The takeaway
If you're redesigning a store with a large or technical catalogue, test the structure before the screens. Card sorting is cheap, fast and honest. It tells you the words your customers use and the way they divide the world, and both of those end up in your search bar whether you designed for them or not.
You can read the full case study, including the research, personas and screens, on my portfolio: The Sustainable Sequin Company case study.
I'm a product designer based in Brighton with a computer science engineering background, so I research, design and build. I'm open to product and UX designer roles (full-time, part-time or contract) and to freelance projects, especially e-commerce and Shopify. Get in touch.
