loading craft.pkg
0%
Back to work
03Product Design × Research

Redesign Navigation

Spendflo's top navigation hit a ceiling as features multiplied within each module. Audited, tested, and validated a left-nav information architecture that cut rage clicks 45% and quick backs 86%.

Role
Product Designer
Timeline
5 days
Tools

Context

A navigation pattern that hit a ceiling.

Spendflo is an AI-native procurement and SaaS spend management platform designed to automate and optimize B2B software purchasing. It centralizes contracts, tracks renewals, and provides AI-driven negotiation and vendor management.

About Spendflo's navigation

The top navigation housed four core modules. It worked early on, but as features multiplied within each one, the pattern hit a ceiling. I took up the initiative to solve it.

Top navigation

Requests

Raise and track purchases

Vendors

Every supplier, one record

Contracts

Agreements and renewals

Insights

Spend analytics

Audit: The Problem

The top nav was cluttered, brittle, and out of room.

Cluttered“There's so much in the nav, I can never find what I need without hunting.”
Procurement manager
Not responsive“On my tablet the navbar buttons slide off-screen, so I can't reach key actions.”
Finance approver
Deep nesting“Settings are layers deep, and some pages hold just one setting.”
Workspace admin
No room to search“There's no global search, so I click through every module to find things.”
IT lead

The old top nav had no room left for global search.

Settings: a flat hub hiding deeper, single-setting subpages.

Research

Watching users hunt for features they should've found instantly.

Usability testing

I conducted task-based testing with users, covering tasks like accessing Requests and switching between modules. I tracked two things: time to locate features, and confusion or drop-off points.

trial and error,

no clear path

needed guiding

no global search,

flagged by users

on their own

Behavioral data

To cross-check what testing surfaced, I filtered for power users and sessions with high navigation changes, then reviewed recordings, click patterns, and heatmaps in Microsoft Clarity.

45% struggled

to find menu items,

paused, backtracked

rage clicks 20%,

quick backs 35%,

all around the nav

Microsoft Clarity heatmap: quick-back-filtered sessions clustered on Requests and Vendor Management, the two most-needed nav items

Information Architecture

Mapping the current IA before designing the next one.

I mapped the existing IA to get a clear picture of what was there: every feature, where it lived, and how it was grouped.

I explored multiple IA directions to check what fits best. One thing became clear early: left nav was the industry standard for a reason. It solved for scalability, hierarchy, and screen real estate in ways the top nav simply couldn't.

Current IA: top nav, hub-like settings, not scalable
Suggested IA: left nav, single settings view, scalable

Visual Directions

Three directions, evaluated.

The first two weren't dead ends, they were learnings: each one ruled out a failure mode and narrowed what the real answer had to do. That's how I arrived at Direction 3.

Direction 03 / 03Shipped

03A nav that gets out of the wayShipped

Instead of collapsing to icons, remove the rail entirely on demand. Full focus, and the smaller-screen problem disappeared as a side effect.

Solution

Left nav, grouped by intent, one click into settings.

Direction 3 shipped: a left nav, grouped by intent, that can hide when you need to focus.

Why Direction 3 was the best fit

  • Focus mode: hiding the nav gives the page back its screen space.
  • Smaller screens, solved for free by that same hidden state.
  • Room to grow: new modules slot in, where the top nav had run out.

What changed, and why

12345
  1. 1Requests & Home Most-visited destinations, kept separate and accessible.
  2. 2Grouped by intent Related actions sit closer together.
  3. 3Accordions Focus on one section at a time.
  4. 4One-click settings Rare settings, consolidated and out of the way.
  5. 5Contextual top nav Only page-level actions remain, no structural nav.

Settings, consolidated into a single view

123
  1. 1One click back A single exit point back to the main product, not buried settings-within-settings.
  2. 2Every setting, one list No more dedicated pages for a single toggle, everything lives in one scrollable sidebar.
  3. 3Contextual help Ask Flo is available right where configuration gets confusing.

Global search, finally given room to exist

123
  1. 1Search trigger Lives in the header now, not squeezed into the old top nav.
  2. 2Scoped filters Narrow results by Requests, Vendors, Contracts, Settings, or Users.
  3. 3Recent searches Picks up where you left off, no retyping.

Notifications, opened from the header

123
  1. 1Notification bell One click from anywhere in the product.
  2. 2Filter by relevance All, mentioned, or assigned to you.
  3. 3Mentions & tasks Surfaces overdue work and direct mentions first.

Using AI tools to move faster

Claude

A design review agent weighed pros and cons on each iteration and built a prototype for PMs, cutting the back-and-forth.

GPT

A copy-review pass on the nav menu labels.

Result and Impact

Shipped, measured, and it worked.

The toggle in motion: nav visible to nav hidden, same screen

I handed off the designs to developers and guided them with prototypes. The impact post-development was immediate.

  • One extra click removed across all navigation layers.
  • No more menu-item clutter: everything moved to the left nav, and it operated well on smaller screens.
  • Easier to find menu items with grouping by intent.
  • Scales well with new feature additions like global search and the Chat with Us experience.

After a week's observation, users backed out of screens far less often, the clearest signal the new IA matched how they actually looked for things.

86%
Quick backs
45%
Rage clicks
52%
Time to locate a feature
4/5
NPS