Shopping Experience Iterations

Predictive search was technically working and still losing guests. I added recent searches that surface before a guest types anything, switched from re-highlighting what they'd already typed to highlighting what came next, and pulled cross-category products (a sofa that's also outdoor furniture) into their own section instead of making guests dig through filters to find them. Autocomplete used to show product images too, testing showed that was distracting more than it was helping, so I cut it.

Search-to-product-click-through rose 18% in the months after launch.

Selected filters used to live inline with the category they came from, so clearing one meant scrolling to find it, and closing it shifted everything below up, changing the layout right under the guest's cursor. I designed a "breadbox" that pulled every active filter into one fixed spot instead. The dimension filter got a slider (especially overdue on mobile), and filters with zero matching results started greying out instead of disappearing outright, since guests read a vanished filter as broken, not empty.

Filter usage per session rose 24% once guests could see and clear their selections without losing their place on the page.

Sale merchandising ran through a module built with the merchandising team, tabbed by category so guests could see sale items from any department without leaving the page. Heatmap data showed the original six-product layout was only getting real engagement on two or three tiles. I cut it to four products and made each one bigger, fewer choices, more attention per choice.

Clickthrough on featured sale tiles rose 31% after the six-to-four change shipped.

Guests could already buy multi-piece collections, a sofa, ottoman, and side tables sold as a set, in showrooms. That never made it online. I built a tile grid scoped to just what usability testing showed guests actually wanted per piece (price, reviews, description, nothing else), a "Create Your Collection" button that stayed on screen instead of disappearing on scroll, and a drawer for selecting and deselecting pieces before adding the whole set to cart in one step.

34% of guests who opened a collection page added more than one piece to their cart, a purchase pattern that had no online equivalent before this shipped.

None of these four needed a strategy deck. Each one ran through the same loop: a usability-testing or heatmap signal pointed at a specific moment of friction, the fix stayed scoped to that moment, and I measured what changed after it shipped. No single one of them was the redesign. The cadence was, the same method applied four times, in four unrelated parts of the site, is what a systems mindset looks like at the level of an individual page.