How to design ecommerce navigation and filters

Begin with customer choices

Study how customers describe products, narrow options, compare alternatives, and recover when they do not know the exact name. Use search queries, support questions, sales conversations, returns reasons, merchandising knowledge, and catalogue data to identify the attributes that matter.

Internal departments and supplier taxonomies may help operations but often make poor customer navigation. A shopper should not need to know how the business stores inventory in order to find the right product.

Build categories with clear boundaries

Give each category a distinct purpose, recognizable label, and useful product set. Avoid placing the same products everywhere unless multiple paths reflect genuine buying behavior. Test whether customers can predict what they will find before opening a label.

Plan landing content for important categories without pushing products too far down the page. Explain meaningful differences, then let people browse. The guide to ecommerce product content shows how consistent data supports discovery and comparison.

Offer filters that change the decision

Prioritize attributes customers use to rule products in or out: compatibility, dimensions, material, use case, availability, price, or other category-specific facts. Do not expose every database field. A long filter panel full of weak choices creates work without improving decisions.

Use understandable values, consistent units, useful ranges, visible selections, result counts, and an easy way to clear filters. Decide how unavailable combinations behave. On mobile, keep the current result context and make the number of applied filters obvious.

Connect search to catalogue language

Support product names, categories, important attributes, common synonyms, model numbers, and frequent misspellings where evidence justifies them. Results should explain why an item matched and provide useful recovery for zero-result searches.

Search and navigation should converge on the same reliable product data. If marketing copy says “weatherproof” but the catalogue only stores an undocumented code, neither filters nor search can use the claim safely. Fix the data ownership rather than layering interface workarounds.

Test discovery with products and tasks

• Tasks: representative goals, not instructions using menu labels.
• Catalogue: common, edge-case, unavailable, and incomplete products.
• Devices: touch, keyboard, small screens, and large screens.
• States: no filters, many filters, zero results, and returning visits.
• Metrics: product views, refinements, zero results, and downstream orders.
• Ownership: people responsible for taxonomy, synonyms, and attributes.