There is a specific failure that every merchandiser has watched happen and almost nobody has a name for. A shopper lands on a category, sees a product attribute printed on the card, tries to filter by it, and finds the filter is not there. They were shown the fact and then denied the ability to use it.
Baymard Institute puts a number on it. In their product list and filtering research, 38% of sites fail to provide filters for information they already display on the list item, which they identify as a direct cause of abandonment. This is not a survey of small merchants: their 2025 benchmark covers 21,000+ manually reviewed parameters across 170+ leading ecommerce sites in the US and Europe, built on more than 200,000 hours of usability research.
Sites that display an attribute and then cannot filter on it
Baymard Institute, product list and filtering benchmarkBaymard identify this as a direct cause of abandonment. The benchmark covers 21,000+ manually reviewed parameters across 170+ leading US and European retailers.
View as table
| Share | |
|---|---|
| of tested sites offer no filter for information they already display | 38% |
| offer filters for what they display | 62% |
Why this is a data problem, not a design one
The instinct is to treat it as a front-end oversight. Somebody forgot to add the filter. Occasionally that is true, and it is a good afternoon's work when it is.
More often the filter is missing because it could not be built. A filter needs a structured value on every product in the set, and what is printed on the card is frequently derived from a description rather than read from a field. The fact is visible because a template rendered a sentence. It is not filterable because there is nothing to compare.
The attribute is on the page and not in the schema. A human reads it. A filter cannot.
The measurement problem underneath
Say you accept the diagnosis and decide to fix it. Which attribute do you structure first?
The usual answer is to rank by how empty each field is, because that is the number a completeness report gives you. That ranking is close to arbitrary for this purpose. A field that is 90% empty and that nobody chooses on is worth less than a field that is 40% empty and decides every purchase in the category.
On one laptop shelf we measured ourselves, ranking by absence and ranking by what shoppers said they needed disagreed completely. Processor was missing on 162 of 216 products, which put it third by absence. Only 2 of 32 stated buying requirements mentioned a processor at all, which put it last by demand. Storage and processor traded places, and fixing processor first would have been the most expensive way to change nothing.
What to do about it this quarter
- Open your worst-performing category as a shopper, not as an admin, and try to narrow it the way a buyer would
- List the attributes printed on the card that you cannot filter on. That list is usually shorter and more actionable than a completeness report
- Rank that list by whether a purchase decision turns on it, not by how empty it is
- Check how much of it is recoverable from your own product titles before you budget for acquisition
That last point is the one that changes project sizing. On the shelf we measured, 161 of 216 missing values were sitting in the merchant's own product titles, recoverable with a script. A quarter-long enrichment programme and a week of extraction look identical on a completeness dashboard and cost very different amounts.