Facet Order Is Merchandising. Most Stores Sort It Alphabetically and Call It Done.

Which filters appear, in what order, with which values first, decides whether shoppers narrow a category or abandon it. Alphabetical is an abdication, and shoppers read it as one.

A category page with 800 products is not a page, it is a question, and facets are how the shopper answers it. Which facets exist, the order they are presented in, and the order of values inside each one are merchandising decisions as real as the order of products on the page, and most stores make them by accident: whatever the platform generated, sorted alphabetically, identical on every category. Baymard's filtering research keeps finding the same shape of failure. Shoppers need category specific facets, the attribute that actually differentiates products in this category, presented early, and instead they get a generic stack where the deciding attribute for shoes sits below three facets nobody uses, with values ordered by the alphabet rather than by what shoppers pick. A shopper who cannot narrow 800 products to a dozen in a few clicks does not have a discovery problem. The store has a sorting problem, and the shopper solves it by leaving for a store whose filters work like a salesperson: the important question first.

Lead with the deciding attribute per category

Every category has one or two attributes that carry the decision: waterproofing and width in trail shoes, capacity and door style in refrigerators, shade family in foundation. That attribute should be the first facet, expanded, with its values ordered by selection frequency or by the natural scale of the attribute, sizes ascending, prices in bands that match how budgets actually cluster. Generic facets, brand, price, availability, belong in the stack, but after the deciding attribute, because the shopper's mental task is narrowing by what differentiates, and the interface should follow the task.

Value order inside a facet is its own quiet lever. Alphabetical size lists that interleave numeric and letter systems, or color lists that lead with the three shades nobody buys, add friction to every single narrowing action. Order by popularity within a sensible structure, keep out of stock values visible but marked where shoppers plan around availability, and never let a facet value with zero current products sit there as a trap that empties the page.

Related reading on this site: RelevanceRail and Zero Results Is a Ranking Failure Wearing a Search Costume cover the neighbouring problems in more depth.

Too many facets is a failure mode too

The opposite sin is the twenty facet rail, every attribute in the database exposed because it could be filtered on. Shoppers do not experience that as power. They experience it as homework, and Baymard's testing consistently shows long undifferentiated facet lists going unused below the first few entries. Curate ruthlessly: five to eight facets per category, chosen from actual narrowing behavior, with the long tail of attributes left to search. Retire facets nobody selects. A facet chosen by under one percent of category visitors for a quarter is not serving shoppers, it is serving the database schema's feelings.

Mobile raises the stakes because facets hide behind a filter button and every open is a commitment. On a phone the first screen of the filter panel is the whole game: deciding attribute first, common refinements visible without scrolling, applied filters shown as removable chips so the shopper can see and undo their own narrowing. A shopper who cannot tell which filters produced an empty result set will blame the store's stock, not the interface, and leave with the blame attached to the brand.

Measure facets like you measure products

Facet analytics are the neglected half of category reporting. Track selection rate per facet, the rate at which a facet's use ends in a purchase versus an abandoned narrowing, and the dead end rate: facet combinations that return zero products and strand the shopper. Dead ends deserve a designed answer, the nearest relaxation suggested in plain words, remove the size constraint to see 14 more, and never a blank page with a reset button as the only thought the store had. Each quarter, promote the facet shoppers keep reaching for through search terms, demote the ones gathering dust, and the category page slowly becomes what it should have been at launch: an argument, in the shopper's order, for the right dozen products.

Frequently asked questions

What order should facets appear in on a category page?

The deciding attribute for that category first, the characteristic that actually differentiates the products, followed by common refinements like brand, price, and availability. Value order inside each facet should follow selection frequency or the attribute's natural scale, not the alphabet.

How many facets should a category page have?

Usually five to eight well chosen facets. Long generic lists go mostly unused and turn narrowing into homework. Facets selected by almost no visitors over a quarter should be retired, leaving deep attributes to search.

What should happen when filter combinations return no products?

The page should suggest the nearest relaxation, such as which constraint to remove to see more results, rather than showing a blank result set. Dead ends with no guidance make shoppers assume the store is out of stock and leave.

How can a store tell if its facets are working?

Track per facet selection rates, whether facet use leads to purchase or abandonment, and how often combinations produce zero results. Those metrics show which facets to promote, reorder, or retire each quarter.