Product craft
A navigation redesign reaching tens of millions
How millions of people move across Jira, Confluence, and the rest of the Atlassian suite. A year of planned navigation work landed in three months — without the team’s quality bar slipping under the compression.
- Role
- Design Manager
- Company
- Atlassian
- Timeframe
- 2025 – 2026
- Scope
- Platform navigation
01The scale of the layer
Navigation is the layer nobody fights over because it doesn’t feel like a feature — until it breaks. It decides whether a user who lives in Jira can find their footing in Confluence, whether “where am I?” ever needs asking, and whether a suite feels like one product or six strangers.
Tens of millions
of people cross this layer every working day.
A whole suite
Jira, Confluence, and the products around them — each with its own users and inherited debt.
One pattern language
The output is cohesion: navigation patterns users learn once and recognise everywhere.
02From six wayfinding systems to one
Before — every product its own maze
Each product grew its own navigation — different homes, different sidebars, different mental models. Users paid the tax at every seam.
After — one rail, learned once
One shared rail with consistent product switching, search, and creation — the same grammar wherever you land.
03A year of roadmap, one quarter to land it
Priorities shifted, and the plan compressed: a year of sequenced navigation work had to land in a single quarter. The risk wasn’t missing the date — it was making the date by quietly lowering the bar.
We held it by being explicit about the trade: design reviews stayed, exploration got timeboxed, and anything cut was cut out loud — a named decision with an owner, not an erosion nobody chose.
04Where it landed
Shipped at platform scale, quality bar intact
The compressed plan landed across the suite without the team’s standards slipping — the reviews, critiques, and pattern discipline survived the schedule.
One grammar across products
Product switching, search, and creation now behave the same wherever a user lands.
A team that owned the trade-offs
Cuts were named decisions with owners — the team came out of the compression sharper, not burned.
Full case study in progress. Adoption and satisfaction metrics exist but are being cleared for public use — I’d rather show real numbers late than approximate ones now. Ask me about them in a conversation.