As consumers and business buyers, it’s irritating if “0 results” pages pop up when we’re searching for products. It happens far too often and the symptoms are self-evident: Internal searches feel like blunt knives cutting steak and visitors end up bouncing from the channel. It looks like a UX problem, but actually, the root cause is product data.
Below, we look into the issue of why search filters break, and how better PIM and taxonomy can solve this problem.
When filters fail, it is rarely just the website
Broken filters mean clunky menus, non-intuitive mobile layouts, confusing labels, and the aforementioned “zero results” information cul-de-sacs. These all damage the CX, and we need to know why.
Faceted (search criteria-based) navigation is a window onto how you manage your product information. Filters read the same taxonomy, attributes and variant structures stored in your PIM or catalogue. When these ‘commercial vertebrae’ are insubstantial, inconsistent or poorly governed, the user interface has nothing concrete to work with, resulting in shoppers seeing partial ranges, odd gaps, or nothing at all.
A weak taxonomy: where products go to die
Product taxonomy acts as the filing system behind every filter. If it’s confusing (or confused), product filters will be as well.
The most common fails?
- Inconsistent categorisation
Is a blazer found under the category “jacket”, or “workwear” or “suits”? If similar products are spread across different categories, applying a filter like “jackets” feels random.
- Overly broad or painfully narrow levels
As far as filter categories go, “Electronics” is far too non-specific, while “left-handed 18V brushless SDS drill for concrete” goes absurdly to the opposite extreme. In these cases, you’re either faced with a filter which doesn’t narrow your choice, or a filter which only applies to 2 SKUs.
- Flat hierarchies with no real drill-down
If you leave your offering in only two or three top-level categories, the browser can’t progressively narrow a large catalogue without being overwhelmed by filters which lack suitable context.
Modern PIM platforms are designed precisely to handle hierarchy management and multiple taxonomies, whether by channel, region or customer type, but they can only perform this function when the business user designs and continually monitors that structure.
Attribute chaos: why “0 results” really happens
Essentially, filters work by reading product attributes. If the data is insubstantial, incomplete, or inconsistent, the filter is effectively working blind.
Again, typical issues repeatedly emerge:
- Empty fields
You cannot offer a “screen size” filter if 60% of your televisions have no information on that attribute. Sure, the products are displayed on the page, but they won’t appear if a user narrows using that facet.
- Free-text values everywhere
“Navy”, “navy blue”, “midnight”, “blue/navy”. A human can see the difference, but to a filter, it’s four separate buckets[1]. As a result, customers who choose “navy” only get to see a part of the range you actually sell.
- Attributes buried in prose
Take some key attributes in a B2B distributor’s listing: “IP67”, “12mm shank” or “R10 slip rating”. If these only appear in a paragraph of continuous text rather than in structured fields (like a features bullet list or a table), your filters and internal search engines won’t be able to use them reliably.
This is where product information management and product data enrichment make the difference. A PIM, with strong pre-established data governance, makes key filter attributes mandatory. Thus, teams work with controlled value lists and can use AI-assisted onboarding to harmonise supplier feeds before they ‘contaminate’ the catalogue with substandard data.
The governance gap: Who actually owns filter quality?
When you dig down, you’ll find that behind every broken filter is a governance problem, in that there’s no clear owner for the product data which drives discovery.
Lack of clear data stewardship[2] results in:
- Ad hoc attribute naming (Colour / Colour Name / Product Hue)
- Different teams using their own naming conventions
- No agreed single place to resolve discrepancies / inconsistencies
Industry analyst, Gartner, particularly highlights workflow, data quality and stewardship as core PIM capabilities, not just vague projects for future implementation. For a PIM to work to best effect, its filters should be rules-based – for instance:
- Which attributes can surface as facets
- What the “good enough to publish” threshold means for each one
- Where (or with whom) does the buck stop if something goes wrong
At Start with Data, we very much typically frame this issue as a business problem, not an exercise for your IT to sort out. It’s very much about designing operating models, roles, and processes around PIM. Once they are in place, taxonomy and filters stay aligned with how customers actually shop, not how systems were configured five years ago.
Confused by PIM Vendors?
With 100s of PIM software vendors worldwide, choosing the right PIM solution can be a daunting & confusing task.
Use our guide to assess PIM solutions against the right capabilities to make an objective and informed choice.
Variants, channels and the ‘invisible catalogue’
A characteristic area where filters cause most problems is that of complex products. Such as fashion items and accessories, configurable B2B items with minimal variants, or multi-channel ranges.
The two common pitfalls in this case are:
- Parent–child confusion
Colour and size might live on the variant SKU, while search and filters only index the parent. Someone filters for “blue, size 12” and half the relevant items never appear because the system only reads “dress” on the parent.
- Channel-specific requirements
Marketplaces enforce their own taxonomies and obligatory attributes. If your internal model is too loosely defined, mapping to “proper” category and filter structures becomes an organisational pain in the neck, especially when doing it at scale – think of marketplaces like Amazon, or specialist distributors or large B2B portals.
Fortunately, modern, composable PIMs support rich variant models and multiple taxonomies per channel. However, your business will only benefit if you treat product data management as an asset:
- Designed
- Governed
- Continually improved
Turning your broken filters into a discovery engine
Fixing failing filters with a JavaScript widget is a sticking plaster solution. A truly long-term fix is to treat your product data as infrastructure. A practical roadmap:
- Run a data health check on your key filters
Pick your highest-value categories and audit completion, consistency and value lists for every facet.
- Define a “golden” attribute set for filtering
Decide which attributes genuinely help users choose (by category) and formalise those as canonical filter fields in your PIM or ecommerce site.
- Tighten governance and workflows
Use PIM validation, workflows and AI-assisted enrichment to block incomplete records and normalise values at the point of onboarding, not after complaints arise.
- Analytics closes the loop
Use digital shelf analytics, on-site search data and filter usage to see where customers hit dead ends and feed insights back into taxonomy and attribute design.
Get it right and filters will no longer be a CX drag factor but one of the sharpest tools in your kit for product discovery, customer satisfaction and conversion.
Final words
If your filters are failing, it isn’t a UX problem. It’s a product data problem.
Fix the taxonomy, fix the attributes and fix the rules underneath the page, and your filters start behaving the way customers expect.
If you want support getting there, we can help. At Start with Data we focus on repairing the foundations: auditing your data model, tightening your attribute sets, and building governance that actually sticks. When the structure is right, every channel becomes easier to run.
If you want to talk about improving your filters, just get in touch.