Overview
Cisco CrossWorks is a Traffic Engineering platform for ISPs and CSPs that monitors network traffic, predicts issues, and dynamically routes packets using MPLS technology. Device onboarding — the process of registering routers into the network manager — had become a major pain point, described universally as tedious. I owned the Device Inventory and Automation modules across two release cycles dedicated to solving this.
The Platform
CrossWorks comprises five modules: Traffic Engineering, Topology, Device Inventory, Security, and Automation. Onboarding a new router to the network manager is the entry point for everything else — if it's broken, nothing downstream works well.
Understanding the Problem
Onboarding a new router to CrossWorks should be routine. In practice, users were required to enter an excessive amount of information manually, with little guidance on what was actually needed versus what could be inferred. Credentials had to be re-entered for every device. Bulk uploads via CSV demanded 32 separate fields — most of which were optional or could be derived automatically. Errors during upload provided no inline feedback, requiring users to fix issues and re-upload from scratch.
Research Approach
We used a multi-method research approach. Unmoderated usability testing via Maze with first-time internal and external users across geographies gave us behavioural data at scale. Stakeholder interviews with PMs, TMEs, and field executives surfaced the business context and competitive pressure. Competitor analysis of Juniper Paragon and Zabbix revealed patterns that made onboarding significantly less painful elsewhere.
Competitive Analysis
Juniper Paragon and Zabbix both offered onboarding experiences that required significantly less from users upfront. Side panels outperformed modal windows for navigation continuity, auto-discovery was a table-stakes differentiator, and minimising required fields dramatically improved adoption rates.
Design Solutions
Simplified single-device add flow: Redesigned to accept just a device identifier — IP address or hostname — and auto-fetch available information. Users no longer needed to know or manually enter data the system could discover itself.
Bulk Upload: 32 Fields → 8
The CSV template was the single biggest barrier for bulk onboarding. Reducing from 32 to 8 required fields made the template approachable. Real-time verification with inline error correction replaced the upload-fail-fix-re-upload loop. Credential profiles let users define reusable credentials once, referencing them across hundreds of devices.
Enhanced Device Details & Auto-Discovery
Device details were redesigned with visual port representation using chassis imagery, giving engineers a spatial orientation. Configuration backup and rollback capabilities were added. For auto-discovery, users set IP ranges for automatic device detection — closing the gap identified in competitor analysis.
“Onboarding a new router to a network manager is a tedious task.”
Outcomes & Impact
Bulk upload CSV reduced from 32 required fields to 8, dramatically lowering the barrier for large-scale device onboarding.
Single device add, bulk CSV upload, auto-discovery, credential profiles, and ZTP support — all redesigned across two release cycles.
Received positive feedback from both internal teams and customers; detailed comparative metrics tracked as a follow-on UX story.
