Navigation restructure
- My role
- Product design lead (sole designer)
- Team
- 1 product designer, 1 front-end lead, engineering
- Timeline
- 2025
Description
Rebuilding the information architecture of a multi-layered enterprise product so users can move between levels the way they actually think about them.
Read case study
The problem
The product exposes information at three levels: a portfolio of systems, an individual system, and the underlying code. Software architects, the primary users, don’t sit at one level. Their real work is constantly switching between them, zooming out to compare systems and zooming back in to inspect one.
The existing horizontal navigation couldn’t hold that structure. It had run out of room, so features had been progressively buried in drop-down menus. The two levels users switched between most, portfolio and system, shared a similar layout with no clear indication of which one you were in. The nav had stopped being a map of the product and become a cabinet of hidden drawers.
Left alone, this would have kept getting worse: the roadmap had new pages queued up, and there was nowhere to put them.
What I set out to change
- Make the level a user is on (portfolio, system or code) unambiguous at a glance.
- Make deep-dive switching, the dominant behavior, smooth instead of disorienting.
- Surface features that were under-used because they were buried, not because users didn’t need them.
- Create structural room for the upcoming roadmap without another redesign in six months.
Process
Quantitative: what were users actually doing?
I analyzed usage data to identify the most common flows through the product. Two things stood out.
Users switched between portfolio-level and system-level views far more often than the existing IA assumed. The nav treated these as separate destinations; users treated them as two ends of the same task.
Several features the team believed were high-value showed low engagement. When I mapped the click paths, they were consistently three or more levels deep inside drop-downs.
Qualitative: why were they doing it that way?
I ran interviews with five-plus internal stakeholders across roles, weighted toward customer-facing ones who watch real users work with the product. I was looking for the gap between how we assumed the product got used and how it actually got used.
The gap was significant. Features we thought were part of a user’s weekly rhythm were, in practice, hard to find. That’s not a user problem. It’s a navigation problem masquerading as an adoption problem.
Synthesis
I used affinity mapping to consolidate the interview and data findings into a set of design directions. The most consequential decision was structural, not visual: I organized the IA around tasks-to-be-done rather than personas. The product has a few well-defined roles, but different roles do very similar jobs inside it, and the boundaries between who uses what were fuzzy enough that a persona-based IA would have created artificial separations.
From there, the shape of the solution became clear:
- A persistent, vertical left-hand sidebar. Vertical navigation preserves context as users move deep into hierarchical data, and it has room to grow that a horizontal bar doesn’t.
- A flatter structure. Features hidden inside drop-downs come up to the top level and are visible at rest, not on hover.
- Clear location cues, so portfolio and system context is obvious even when the two views’ content looks similar. Pages had no explanatory title for each section, so users didn’t immediately know where they were.
- Fast access to the system overview, which mattered most.
- Deprecating pages that no longer fit the product vision, rather than carrying them forward.
Design and validation
I mapped the current IA end to end, then mapped the proposed one against it to make sure nothing critical was lost in the flattening. Wireframes became an interactive prototype, which I tested with two audiences: existing users, to check that muscle memory wasn’t broken destructively, and users new to the product, to check that the structure was legible without prior context.
Outcome
- Flatter access. A feature that moved from two clicks to one saw traffic rise around 10% in the period after launch. I treat this as a directional signal rather than proof. It’s one page and one timeframe, and it coincided with a broader refresh, but it is consistent with the hypothesis that flatter access reduces friction.
- Room for the roadmap. Shortly after launch, a strategically important new dashboard needed a prominent home. The new navigation had room for it at the top level, without reworking anything around it. That dashboard saw roughly 10% higher usage than a comparable earlier launch, which I read as a sign the placement worked, not only that the redesign made room for it. The navigation was built for the roadmap, not just for the moment it shipped.
- Less visual weight. The vertical sidebar carries the hierarchy on its own, so the main canvas can quiet down. People spend more attention on the page’s content and less on working out where they are.
The new structure also held up under real roadmap pressure. Shortly after launch, a strategically important new dashboard needed a prominent home in the product. The redesigned nav had room to give it one, at the top level rather than tucked into a submenu, without any rework of the surrounding structure. That is the more meaningful result: the nav was built for the roadmap, not just for the moment it shipped.
The chrome also carries less visual weight. The vertical sidebar holds the hierarchy on its own, so the main canvas can quiet down and users spend more attention on the content of the page.