The 10 nav and search patterns in this slice of the vault, the rubric I use to judge a header fast, and the Claude build that turns your search logs into a ranked fix list.
Most teams ship navigation once and never touch it again. Search usually gets bolted on with an app and then ignored. On a small catalog you can get away with both. At 8 figures with a deep catalog you cannot, because the header and the search box are where a shopper's intent either gets routed to the right page or quietly dies. The people typing into your search bar already know what they want. They told you, in their own words. Leaving that surface on autopilot means handing your highest-intent traffic the worst experience on the site.
This page is the teaching layer for the Navigation & Search slice of the vault. Read the rubric, then work the gallery and take the moves that fit your catalog.
A note on numbers before we start: I don't have an audited benchmark for how much better search users convert versus browsers, so I'm not going to quote one. In my own programs the gap is real and large, but treat that as an operator's claim to test on your own data, not a stat.
Five things. The first four are nav. The fifth is search, and the fifth is where most large stores have the biggest untapped lever.
Keep the top level to 3 to 5 items and do the real work in the dropdown. Allbirds runs three top items (Men, Women, Sale) and routes everything else through a gender dropdown ordered footwear, then type, then favorites, then apparel. That order is a decision, not an accident. Most-clicked path first.
The failure mode at scale is the reverse: a top nav with nine or eleven items because every category lead lobbied to get their thing into the header. Every item you add to the top level dilutes the click probability of the ones that matter, and it pushes the priority decision onto the shopper instead of onto your IA. Decide the order yourself, in your data, and let the top level carry only the highest-intent entry points.
The best menus route the way shoppers think, which is by use case or by who they are, not by your internal product taxonomy. Lululemon leads with activity (run, train, yoga) before product type. Gymshark routes gender, then activity, then collection. A shopper arrives thinking "I run" or "I lift," not "I would like to browse the long-sleeve technical tops SKU family."
When your nav mirrors your merchandising spreadsheet, you make the shopper do the translation. When it mirrors their intent, you do it for them. On a big apparel or supplement catalog that is the difference between a menu people use and a menu people scroll past on the way to the search box.
For catalogs people graze rather than search, imagery in the dropdown earns its space. Warby Parker's eyewear mega menu sorts by shape and use and shows the frames. Rothy's runs image-led dropdowns with material and sustainability cues in the nav itself. Recess uses a two-level slide menu with flavor icons so you scan by picture instead of reading a list.
The rule: if the buying decision is visual (apparel, eyewear, anything where look or flavor or finish drives the choice), show the thing in the menu. If the decision is spec-driven, a clean text menu routes faster. Match the menu format to how the category actually gets chosen.
A short catalog does not need a mega menu, and forcing one signals insecurity. Aesop runs a restrained editorial nav that routes a focused range and reinforces the premium position by what it leaves out. The lesson is not that everyone should minimize. It is that nav complexity should match catalog complexity. A 40-SKU brand wearing a department-store mega menu looks like it is pretending to be bigger than it is.
This is the one most operators under-invest in. The shoppers using your search bar typed the exact thing they want, so they arrive pre-qualified. That is also exactly why a bad search experience is expensive: you are failing the people closest to buying. Sephora is the common reference for faceted search at scale, and Nordstrom for predictive search across a deep multi-brand catalog. Both make a catalog too large to browse actually findable. Note that both are large retailers rather than pure DTC, so treat them as the standard your search is measured against, not as same-size peers.