Booking.com

Localization Routing UI

TL;DR

Problem: Routing rules lived in a database, so every tweak meant opening an engineering ticket.

My move: Designed a self-service interface with drag-and-drop logic, real-time validation, and a clear job dashboard.

Impact: Operations teams now edit rules without dev help; translation requests move faster and satisfaction scores rose in post-launch surveys.

Design of Booking.com's Localization Team Routing UI

Timebound: 6 months

Role: Product / UX Designer

Tools: Figma, Miro, Whiteboard

Note: Due to confidentiality agreements, specific details about the platform and proprietary design elements cannot be shared.

Where we were guessing

We knew localization project managers (LPMs) were stuck waiting on developers.

Unknowns on day one:

  • Which decisions needed daily edits, and which could stay hard-coded?

  • How much “technical” language could LPMs comfortably use?

  • How do we create a logic dashboard for such a complex system?

Gray areas

LPMs kept side-spreadsheets to track job status

Ops Slack was full of “quick rule-change” pings

From issues to solutions

What I did with it


Built a live dashboard: colour-coded streams, SLA heat-map, and sortable queues


Designed a step-by-step rule wizard that let LPMs add, edit, and test routing logic


Non-technical users feared “breaking the tool”

Added real-time validation + drag-and-drop canvas that flagged conflicts inline


Bumps along the way

Localization of long language names broke grid; swapped to language codes with tooltip to ensure clarity.

Wizard needed instant rollback, but backend had no transactions. Collaborated with engineers to add a “draft” mode that saves data in a separate table until users hit Publish.

The first validation was direct and red; pilot teams felt criticized. We softened the tone and used inline tips instead of pop-ups.

Leading through the process

  • Deep-dive week on routing UIs: I mapped patterns from workflow engines (HubSpot, Marketo, Salesforce, Zapier) to understand how they work, then distilled the findings into three design rules we referenced all project long.

  • Ran a daily 15-minute huddle with engineering and PM so any rule-builder blockers were surfaced and fixed quickly.

  • Shadowed two localization managers during a live rollout; their on-the-spot feedback fed straight into the next sprint’s backlog.

Impact

What shipped & why it matters

  • Self-service rule builder replaced developer built scripts: ops edits went from 3-day ticket to same-day change.

  • Visual dashboard lets LPMs spot stuck jobs early, cutting follow-up pings to dev by half.

  • Scalable templates handle new clients without extra code: engineering confirmed zero regressions after launch.

Reflection

LPMs didn’t want more control - they wanted safe control. By giving clear feedback, hiding risky actions behind confirmations, and adding escape options (like a break-inheritance toggle), non-technical users could work quickly without fear of mistakes.

The main challenge was bridging two worlds. At the first tech kickoff, engineers talked in technical terms like "regex" and "null-safe strings," while I was still figuring out user flows. Realizing I wasn’t fully understanding the side effects, I involved the PM in every design-dev meeting. Their clear explanations kept decisions focused on user needs, not just coding.

Giving operations a tool they can adjust without a developer’s help frees engineers for other problems and keeps localization moving. Small wins, but with big impact over time.

Previous
Previous

Booking.com Content Guidelines and Reporting

Next
Next

Verto.sh