I led the end-to-end redesign of Huckberry's global navigation, from identifying the underlying systems problem through design and launch. The work connected two sides of the experience: the internal tool merchandising uses to build and curate navigation, and the browsing experience customers use across desktop and mobile.
A connected navigation system that inherits its structure from the product hierarchy while preserving merchandising’s ability to feature campaigns, stories, seasonal moments, and other curated content.
We’re most excited for the functional unlocks for the merchandising team, as well as the consistency between the desktop and mobile experience.
As Huckberry's assortment expanded, its navigation became increasingly difficult to maintain and increasingly limited in what it could expose. The customer-facing symptoms were visible: inconsistent categories, gaps in catalog coverage, and navigation that no longer represented the breadth of what Huckberry sold. Behind those symptoms was a structural problem.
Product categories and classes were managed in the Hierarchy Builder, while navigation was recreated independently in a separate admin. Two sources of truth had quietly emerged.
Routine changes required merchandising to work across two systems with two workflows, with nothing ensuring that the categories customers saw in navigation matched the catalog behind them. Over time, the two inevitably drifted apart.
Making navigation an output of the product hierarchy was the obvious structural fix, but it introduced a different problem. Merchandising doesn't only organize products. The team uses navigation to feature seasonal collections, campaigns, editorial stories, and timely moments — things that intentionally don't belong in the permanent catalog hierarchy. Automating everything would solve consistency by removing the flexibility that made navigation useful.
A navigation that accurately represents the store behind it.
Every category should be discoverable, and the structure customers browse should match the pages they reach.
One structure to maintain without losing the ability to curate.
Merchandising needed to feature campaigns, stories, and seasonal moments without changing the underlying catalog.
How might we make navigation an output of the product hierarchy, without taking away the team's ability to curate it?
Automate the structure. Preserve the point of view.
Navigation would inherit catalog structure from the existing product hierarchy rather than recreating it. What remained manual would be intentionally manual: what merchandising wanted to feature, and when.
One source of truth
Editors search the live product hierarchy and pull categories directly into navigation. The underlying structure comes with them, so nothing needs to be recreated by hand and nothing can quietly fall out of sync.
Campaigns, seasonal moments, editorial stories, and featured content can still be created directly within navigation, because what to feature is a decision, not a data structure. That work doesn't disappear. It simply stops competing with the work of keeping the catalog structure accurate.
Once the underlying model was established, the next question was how merchandising should actually interact with it. My initial instinct was pragmatic: reuse as much of the existing admin pattern as possible and keep the build light.
Two approaches revealed a trade-off between build simplicity for us and spatial clarity for the people using it every day.
The structure teams already used to organize and reorder navigation items, but assembling even a straightforward menu meant stepping through more actions than the task warranted.
Laid out the way the navigation actually appears on the site, so administrators could see the final experience as they worked: moving between sections, inspecting each one, holding several in view without losing their place.
Sometimes the interface that takes more of our time takes less of theirs.
The better idea came through collaboration with engineering. A developer suggested arranging the editor horizontally to mirror the storefront navigation. It required more work to build, but eliminated something the table-based approach couldn't: merchandisers no longer had to imagine what they were creating. The admin and customer experience became two views of the same object.
With the underlying model and editor established, the storefront needed to make a much broader catalog easier to explore without making navigation feel overwhelming. Exploration and discovery are central to how customers use Huckberry, so early concepts leaned heavily into curated content. For launch, we pulled that back. The simpler structure gave the catalog room to breathe while still leaving intentional places for merchandising and editorial content.
The horizontal editor reflects the structure of the live navigation, making the relationship between what merchandising builds and what customers see immediately understandable.
A shared set of components supports core shopping, curated discovery, and editorial content without requiring a new design or build every time merchandising needs something different.
Hover navigation introduced a deceptively small interaction problem. As customers moved diagonally from a top-level category toward its submenu, their cursor could briefly cross a neighboring category and replace the panel before they reached their target. We introduced a short delay when leaving a category, giving the cursor time to travel while keeping the intended panel stable. It's the kind of interaction nobody notices when it works — which is exactly the point.
The redesigned navigation launched in 2026 alongside Huckberry's rebrand. We're tracking two signals: the share of sessions that use navigation, and how those sessions convert. Both are still maturing, so there's no final performance story yet.
What's already true is structural. Navigation and the catalog are no longer maintained as separate copies of the same information. Merchandising has a purpose-built environment it helped shape along the way, not at the end. Desktop and mobile run on one model.
The earliest signal came off the cuff, in the reactions right after we handed merchandising the keys. The problem looked like an aging menu, but redesigning it alone wouldn't have addressed why it drifted in the first place. The best way to simplify an interface is sometimes to simplify the system behind it.
Problem: Mobile shoppers had to create an account before paying, and drop-off at registration kept surfacing in the data.
Impact: Increased mobile orders by 6.6%, about 50 more a day, without giving up account creation as a business goal.
