# Search facets that help and facets that hurt: merchandising filters shoppers understand | DeftMerch

Source: https://deftmerch.com/blog/2026-10-22-search-facet-merchandising-filters.html

DeftMerch answers


Updated October 22, 2026 · DeftMerch answers

**Short answer:** A facet is a promise: click this and the list gets closer to what you want. When facets work, a shopper narrows three hundred products to twelve in three clicks. When they fail, through irrelevant attributes, overlapping values, empty combinations, and counts that bear no relation to the results, the shopper stops filtering and starts scrolling, and scrolling a large catalogue is where sessions go to die.

## Choose facets from shopping behaviour, not data fields

The catalogue database contains every attribute a supplier provided, and platforms happily turn all of them into filters. The result is facets for packaging type, warehouse codes, and minor material variants sitting beside size and colour. Each useless facet taxes attention and pushes the useful ones down the panel, especially on mobile where the filter list is a screen of its own.

The selection test is behavioural: do shoppers in this category actually decide by this attribute, and do they know their value for it before seeing products? Size, price, colour, and key features pass in most apparel and electronics categories. Attributes shoppers cannot self report, like fabric weave or chipset names, belong in product comparison content, not in the filter panel.

## Order and grouping decide whether facets get used

Facet order should follow decision order. Shoppers usually fix the hard constraints first, the size that must fit, the compatibility that must match, the budget ceiling, and then refine by preference. Putting the most used, most decisive facets at the top sounds obvious, yet panels ordered by database field or alphabet routinely bury them under attributes almost nobody selects.

Value grouping matters as much as facet choice. Colour facets fragmented into supplier synonyms, navy, midnight, ink, dark blue, give one colour four weak doors instead of one strong one. Normalising values into the vocabulary shoppers use is unglamorous data work with an outsized effect on filter usage and zero results rates.

## Empty combinations and honest counts

The classic facet failure is the legal path to nothing: each individual selection was offered, the combination returns zero products, and the shopper is blamed with a no results message. Well built faceting recalculates counts after every selection and disables or annotates values that would empty the set, so the interface never offers a door it knows leads to a wall.

Counts must be truthful about what clicking will show. Counts that ignore other active filters, or that update a beat after the results, teach shoppers the numbers lie, and once the panel loses credibility they abandon filtering entirely and with it the fastest route to the right product.

## Pruning is maintenance, not a project

Facet usage analytics show which filters are used, which lead to purchases, and which lead to dead ends. A quarterly prune that removes or demotes unused facets, merges duplicate values, and promotes rising ones keeps the panel matched to how the range and the shoppers evolve. New categories and new product types should trigger the same review when they launch, not a year later.

Finally, test the panel with real tasks. Give five shoppers a product to find with specific constraints and watch where the filters help or mislead. Facet problems are embarrassingly visible in observation and strangely invisible in dashboards, which is why ten minutes of watching beats another report.

Keep reading: [Merchandising out of stock category pages: what to show when the best products are gone](https://deftmerch.com/blog/2026-10-22-merchandising-out-of-stock-category-pages.md), [Search results merchandising: sorting rules that help shoppers find the right product](https://deftmerch.com/blog/2026-10-22-search-results-merchandising-sorting-rules.md), and [Bundles that sell: merchandising product bundles without confusing the shopper](https://deftmerch.com/blog/2026-10-22-bundle-merchandising-product-pages.md) cover the nearby cases in this series.

## Frequently asked questions

### How many facets should a category page offer?

As many as shoppers genuinely decide by in that category, usually a focused set rather than everything in the database. Each extra facet competes for attention with the useful ones, especially on mobile, so selection should follow observed shopping behaviour.

### Why do shoppers hit zero results through facets?

Because the interface offered combinations that cannot coexist in the stock. Good faceting recalculates after every selection and prevents values that would empty the result set, instead of letting shoppers walk into dead ends.

### What is wrong with supplier attribute values in filters?

They fragment shopper concepts into technical synonyms, like four names for dark blue, splitting one useful filter value into several weak ones. Values need normalising into the words shoppers actually use.

### How often should facets be reviewed?

Quarterly as routine maintenance, and whenever a new category or product type launches. Usage data shows which facets help, which are ignored, and which create dead ends, and the panel should be pruned accordingly.