Between 58% and 78% of ecommerce sites have mediocre or worse search UX. That's according to Baymard Institute's 2025 benchmark of over 150 leading sites. Not small players, but major retailers with dedicated UX teams.
So why do so many get it wrong?
Often, it's because the focus lands on what's visible.
The Iceberg
When faceted search fails, the symptoms are obvious. Irrelevant filter options, dead-end "no results" pages, clunky mobile experiences. These are what users complain about, so these are what teams tend to address.
But like an iceberg, the visible problems are just 10% of the story. The real issues sit below the waterline. Without addressing them, UI improvements alone rarely move the needle.
Here's what tends to break search, from the bottom up.
Layer 1: Data (The Foundation)
Faceted search is only as good as the product data behind it. This is the unsexy truth that often gets overlooked.
When suppliers send "Volcano Red" instead of "Red," or dimensions listed as both "230V" and "230 V," those inconsistencies don't just clutter a database. They break facets. Users searching for red products won't find certain items. Colour filters show 47 options instead of 12.
The work here isn't glamorous. Normalising metadata, mapping creative colour names to standard ones, enforcing consistent formatting. It happens far from the UI, which is partly why it gets deprioritised.
But here's what's interesting. Even the most beautifully designed filter panel can't compensate for chaotic data underneath.
Layer 2: The Filter vs Facet Distinction
Most discussions use "filters" and "facets" interchangeably. They're actually different, and the distinction matters.
Filters are broad, static categories that don't change between searches. Think "Men's" or "Women's" in the main navigation. They slice the entire catalogue and apply globally.
Facets are dynamic, contextual attributes that adapt based on what the user is looking at. When someone searches for "running shoes," facets like cushioning level and pronation type become relevant. When they search for "laptops," those same facets would be meaningless. RAM and screen size matter instead.
A common pattern is applying the same generic facets everywhere. Brand, price, and colour are safe defaults, but they're rarely sufficient on their own. A houseplant buyer doesn't care about brand. They care about light requirements and whether it's safe for pets.
When facets are treated as static, users end up scrolling through irrelevant options or missing the ones that actually matter. The UI looks fine, but the experience frustrates.
Layer 3: Device Context
Here's a notable gap. 78% of mobile sites perform mediocre or worse on product list UX, compared to 58% of desktop sites. Mobile isn't just lagging. It's significantly behind.
The core challenge? Desktop faceted search depends on showing filters and results simultaneously. That's how users understand the relationship between their selections and what appears. On mobile, there's rarely room for both.
A common response is hiding filters on a separate screen. Users tap a filter icon, make selections, then tap back to see results. But this removes the instant feedback that makes faceted search useful in the first place.
Better approaches exist. The "tray" overlay pattern shows both filters and results on the same screen. Batch filtering lets users make multiple selections before refreshing. These require intentional mobile design rather than responsive scaling of desktop patterns.
A Useful Mental Model
The iceberg framing offers a useful lens for understanding search UX. Whether building it, auditing it, or evaluating an existing implementation.
Three questions, in order.
- Is the underlying data clean and consistent?
- Are facets dynamic and contextually relevant, or static and generic?
- Was this designed for mobile, or adapted from desktop?
If any of these foundations are shaky, surface-level UI changes tend not to solve the underlying problem.
The 58-78% of sites getting this wrong likely aren't failing because they lack best practices. They may be addressing symptoms instead of causes.
The interesting work often starts below the waterline.