01 The Findability Problem
A store can't sell what shoppers can't find, and at 600 SKUs, finding things is the whole game.
A baby and kids products Shopify store came to us with a complaint we hear in some form from every large-catalog store: "people leave without buying, and then they email us asking if we carry things that are literally on the site." The store stocked over 600 products, from newborn essentials to toddler gear. Conversion sat at 1.0%. Support was answering "do you have..." emails about in-stock products every single day.
Small catalogs forgive bad navigation. Everything is two clicks away no matter how you organize it. Large catalogs punish it brutally. Baymard Institute's ecommerce research has found that a majority of sites have serious category taxonomy problems, and that poorly designed navigation and filtering is one of the main reasons users abandon large-catalog sites without finding products the site actually carries.
The evidence here was everywhere once we looked: search logs full of queries returning nothing, a category tree that mirrored the owner's supplier list instead of how parents think, and filters so generic they went unused. This case study covers the three-part findability redesign (taxonomy, filters, and search) and what it did to a store whose only real problem was that its products were hiding.
The core findability insight: shoppers navigate by their own mental model, not yours. A parent shopping for a two-year-old thinks "toddler, travel, gift under $50." A store organized by brand and product type forces every one of those shoppers to translate, and every translation loses some of them. Navigation design is the craft of matching the store's structure to the shopper's head.
02 What the Findability Audit Found
The audit combined search log analysis, navigation click paths, filter usage data, session recordings, and a card-sorting exercise with actual customers, asking real parents to group the store's products the way that made sense to them. The gaps between how the store was organized and how shoppers think were wide:
TAXONOMY PROBLEMS
Categories organized by supplier and product type ("Feeding Accessories", "Textiles") while parents shop by age and situation
No age-based browsing at all, the single most common way parents shop for kids' products
14 top-level categories, some with 4 products, one with 190
In card sorting, customers' groupings matched the store's categories only ~30% of the time
FILTER PROBLEMS
Only three filters everywhere: price, color, brand, regardless of category
No age filter, no material filter, no "machine washable", the attributes parents actually decide by
Filter usage under 8% of collection page visits; on well-designed large catalogs it's several times that
Applying a filter reloaded the page and scrolled to top, losing the shopper's place every time
SEARCH PROBLEMS
23% of searches returned zero results, and over half of those were for products the store stocked under different words ("binky" vs "pacifier", "onesie" vs "bodysuit")
No typo tolerance: "stroler" found nothing
Zero-result page was a dead end: "No results found" with no suggestions, no popular products, no path forward
Searchers converted at 3x the site average when they found something, which means the most motivated segment was hitting the most broken feature
BEHAVIORAL EVIDENCE
Median products viewed per session: 2.1, in a 600-SKU store
Recordings showed "pogo-sticking": into a category, back out, into another, back out, then exit
Daily "do you carry..." support emails about in-stock items, the clearest possible findability signal
03 The Findability Approach
Findability work has a strict order of operations, because filters and search both depend on the underlying structure being right:
We ran an open card sort with 15 customers: here are 60 representative products, group them however makes sense to you, name the groups. The groupings that emerged became the skeleton of the new taxonomy: age ranges, situations (travel, sleep, mealtime, bath), and gifting. The store's data (search terms, support emails) confirmed the same vocabulary. You cannot guess a mental model from inside the business; the owner had 600 products memorized by supplier.
The category tree defines what filters make sense, and the product data behind both defines what search can match. So the work went: taxonomy and product tagging first (every SKU tagged with age range, situation, material, care), then category-specific filters built on those tags, then search tuned with synonyms from the real query logs. Doing these out of order means redoing them.
Conversion is the lagging indicator. The leading indicators are zero-result search rate, filter usage, products viewed per session, and pogo-sticking in recordings. We baselined all four before changing anything, so every piece of the redesign could be judged on whether it actually helped people find things, not on whether it looked cleaner.
04 Rebuilding the Category Tree
The new navigation gives parents the three doors they actually arrive with, instead of forcing everyone through a product-type warehouse map:
The tagging investment: the unglamorous heart of this project was tagging 600 products with age range, situation, material, and care attributes, about a week of structured work. Every visible improvement (the new categories, the filters, the search results) runs on those tags. Findability is a data problem wearing a design costume.
05 Designing Filters That Get Used
Baymard's filtering research finds most ecommerce sites offer generic filters when shoppers need category-specific ones. The redesign made filters match what a parent is actually weighing in each category:
| Change | What We Did | Why |
|---|---|---|
| Category-specific filters | Sleep gets tog rating and crib size; mealtime gets material and dishwasher-safe; clothing gets size and machine-washable | Generic price/color/brand filters answer questions parents aren't asking. The filter set is the category's buying guide in disguise |
| Age filter everywhere | Every collection filterable by the same four age ranges used in navigation | Age is the universal dimension of this niche. One consistent vocabulary across nav, filters, and product pages |
| No reload, no scroll jump | Filters apply instantly and keep the shopper's scroll position; active filters shown as removable chips | The old reload-and-scroll-to-top punished every filter use. Interaction cost is why the old filters sat unused |
| Result counts on every option | "Machine washable (34)" instead of just "Machine washable" | Counts prevent dead-end zero-result filter combinations and teach the catalog's shape as you browse |
| Mobile filter sheet | Bottom sheet with 48px rows, apply button pinned in the thumb zone | Most traffic is mobile; a filter UI that's fiddly on a phone doesn't exist for most visitors |
06 Fixing On-Site Search
Searchers were the store's most motivated visitors, converting at 3x the site average when search worked, and nearly a quarter of their queries hit a wall. The fixes, in order of impact:
Every zero-result query from six months of logs was mapped to the product it should have found: binky→pacifier, onesie→bodysuit, sippy cup→toddler cup, dummy→pacifier (the store had UK traffic), plus typo tolerance. This is the highest-ROI hour in search design: the customers had already written the dictionary, one query at a time, and nobody had read it.
The search field (now permanently visible, not behind an icon) shows product cards with images and prices after three characters. Most searches now end in a product tap without ever reaching a results page. Fewer steps between "I want" and "there it is" is the entire job.
When search genuinely has nothing, the page now shows near-match suggestions, the age-range categories, and best sellers, with a one-tap "ask us if we carry this" that emails support with the query pre-filled. Those emails now double as a stocking wishlist for the owner. A dead end became a conversation.
07 Results: The Catalog Became Visible
The rebuild shipped in sequence over six weeks: tagging and taxonomy, then filters, then search. Nine weeks after the taxonomy launch, with traffic steady:
| Metric | Before | After (9 Weeks) | Change |
|---|---|---|---|
| Overall conversion rate | 1.0% | 1.6% | ↑60% |
| Zero-result searches | 23% | 7% | ↓68% |
| Products viewed per session | 2.1 | 5.0 | ↑2.4x |
| Filter usage (collection visits) | 8% | 36% | ↑28pts |
| Search usage per session | 9% | 19% | ↑10pts, visible field |
| "Do you carry..." support emails | Daily | A few per month | ↓~85% |
Metrics from GA4, Shopify analytics, search app logs, and the store's helpdesk, comparing the 8 weeks pre-launch against weeks 2–9 post-launch. Traffic within 6% across periods, no campaigns or catalog changes in the window. Average order value also rose 11% as sessions surfaced more products, though that metric moves with product mix and deserves less weight.
Context: the metric that best tells this story isn't conversion. It's products viewed per session going from 2.1 to 5.0. The store didn't get more persuasive. Shoppers simply started seeing the catalog, and a 600-SKU store that shows each visitor five relevant products instead of two converts more by arithmetic alone.
08 Why It Worked
Three decisions carried the result.
The card sort replaced guessing with evidence Biggest lever
Fifteen customers sorting sixty products revealed in an afternoon what three years of internal debate hadn't: parents think in ages and situations, not product types and materials. Every downstream decision (categories, filters, search synonyms) flowed from that one exercise. It cost almost nothing and it's the single most skipped step in navigation redesigns.
The tags did the heavy lifting Data layer
One week of tagging 600 products with age, situation, material, and care powered everything visible: multi-door categories, category-specific filters, and richer search matching. The same product now gets found through navigation by one shopper, filters by another, and search by a third. Structure beneath, three surfaces above.
The customers had already written the search dictionary Listening
Six months of zero-result queries was a complete, free, customer-written list of every vocabulary mismatch in the store. Mapping it to synonyms cut zero-result searches by two thirds and rescued the store's highest-converting visitor segment. Most stores are sitting on this exact list and have never opened it.
09 Key Takeaways for Store Owners
It's a customer-written list of what people want and can't find, including products you stock under different names. Every entry is either a synonym to add or a product to consider stocking.
Each email represents dozens of shoppers who didn't bother asking. Treat support tickets as free UX research; they tell you exactly which products are hiding.
Ten to fifteen real customers grouping your products reveals the mental model your navigation should mirror. It takes an afternoon and prevents reorganizing 600 products around another wrong guess.
Generic price/color/brand filters go unused because they don't match how anyone decides. Filter options are the questions buyers actually weigh, with result counts, applied instantly, without losing scroll position.
Tag products once with the dimensions your shoppers think in, then surface them by every relevant path: age, situation, gift, brand. One shopper's "travel gear" is another's "gifts under $50", and both should find the same cup.
Reading Time: 10 minutes · Category: Ecommerce Design · Navigation UX · Findability
Next case study Product Image System →