Facets that help shoppers narrow instead of drowning them

Watch a shopper in your top category and note the first thing they narrow by. In apparel it is usually size, because a wrong size is a return. In appliances it is often dimensions or capacity, because the thing must fit the space. That first decision belongs at the top of the facet list, expanded, with the long tail of attributes below it or behind a more filters control.

Feed order is the enemy here. Product data arrives in whatever order the supplier chose, and unedited facet lists inherit it. The edit costs an afternoon per major category and it is the whole job: a merchandiser deciding what the shopper decides by, in sequence.

Order facets by the decision, not the database

The number beside a facet value is a promise about what clicking it will show. When counts are stale, cached from yesterday's inventory, or computed before other filters apply, shoppers click into disappointment and learn to distrust the whole sidebar. Dynamic counts that update with each selection are more work and visibly more honest, and the click data will show the difference within a week.

Zero result combinations should not be clickable at all. Grey them out or remove them as other facets are chosen. Every dead end a shopper reaches is a small accusation that the store does not know its own stock, and on mobile, where backtracking costs the scroll position too, dead ends end sessions.

Counts are promises, keep them honest

On a phone the facet list hides behind a button, which means the order matters even more: the first screen of filters is all most shoppers will ever open. Put size, price, and the category's deciding attribute there. Bury nothing essential below a second expansion, because almost nobody expands twice.

Applied filter chips that can be removed individually are the other mobile essential. A shopper who cannot see what is filtering the list, or must reset everything to change one value, will reset the session instead. Show the active filters above the products, each removable, with the result count updating as they go.

Mobile shoppers get a different facet problem

Track facet use to purchase, not facet use alone. A filter that gets clicked constantly and converts rarely is either mislabelled, miscounted, or narrowing to products that disappoint. A filter nobody clicks might be useless, or might be in the wrong position, and moving it up for a fortnight answers that question with data instead of opinion.

Clean up values while you are in there. Suppliers send the same colour as navy, dark blue, and midnight across ranges, and each variant becomes its own facet value with its own thin count. Normalising attribute values is unglamorous data work that doubles the usefulness of the facet it touches.

Measure facets by what they finish

Keep reading: Zero results search pages: the most honest page in your store is wasting the visit, Evergreen gift guides: build the guide once, sell from it all year.