IBM Network Capacity Management — Unifying fragmented infrastructure planning into a single decision platform.

What this project was and what it achieved.

What was broken, why it was hard, and who was affected.

Network capacity planning at IBM was managed through a patchwork of tools — Excel spreadsheets for tracking, emails for requests, multiple vendor systems for data, and no consolidated view of current or future demand. Network operations engineers were spending the majority of their time on manual coordination rather than actual planning. The platform was invisible: when capacity failed or was miscalculated, teams found out too late.

• No single source of truthTeams worked across 4+ disconnected systems with no unified capacity view
• Manual request processingEvery capacity request required email chains, averaging days to resolve
• Zero demand forecastingNo predictive capability; teams reacted to problems rather than anticipating them
• Repetitive allocation tasksHigh-frequency allocation work that could be automated was done manually
• No audit trailRequests were lost, duplicated, or untracked, creating accountability gaps

What I learned from 21 users.

I conducted 21 semi-structured remote user interviews with network operations engineers and capacity planners. The research focused on understanding current workflows, pain points, workarounds, and decision-making processes.

• Insight 01
  The bottleneck was coordination, not capability. Engineers spent 60%+ of their time on emails and status checks, not actual capacity decisions. Workflow consolidation and automated routing would free significant cognitive bandwidth.
• Insight 02
  Users didn't need more data — they needed fewer, better decisions. Dashboards existed but were overwhelming; users had developed personal workarounds to filter noise. Design for decisions, not data display.
• Insight 03
  Forecasting was the highest-value missing capability. Every participant cited reactive capacity failures as their biggest pain point. Demand forecasting was a must-have, not a nice-to-have.
• Insight 04
  Trust in the system required transparency. Users distrusted automated suggestions without understanding the underlying data. Every automated action needed a visible, understandable rationale.

How I translated research into a product direction.

The research pointed to four governing design principles that shaped every subsequent decision:

• One source of truth
  Why

  Users were switching between 4+ tools. Consolidation was the single highest-value intervention.
  Trade-off

  Migrating existing data and workflows was technically complex and required significant stakeholder alignment.
  Impact

  Single unified platform eliminated tool-switching entirely.
• Design for decisions, not data display
  Why

  Dashboards existed but weren't actionable. Users needed answers, not numbers.
  Trade-off

  Required deep collaboration with data engineering to surface the right signals.
  Impact

  Capacity health, demand forecast, and request status became scannable at a glance.
• Automate repetitive allocation
  Why

  67% of allocation tasks were identical. Manual processing was waste.
  Trade-off

  Automation required explicit human override paths to maintain trust and accountability.
  Impact

  80% reduction in manual effort across the planning workflow.
• Progressive disclosure for technical complexity
  Why

  Engineers and managers had very different information needs from the same interface.
  Trade-off

  Required more complex IA and role-based view architecture.
  Impact

  Both technical and non-technical users could navigate the platform confidently.

How IBM Bob AI Agent accelerated design delivery.

I leveraged IBM Bob AI Agent within Figma for design development and GitHub for cross-functional collaboration. This AI-assisted workflow reduced design-to-engineering handoff time significantly and enabled faster iteration cycles. The platform itself also incorporated AI-driven demand forecasting — surfacing predictive capacity insights that were previously impossible with manual data aggregation.

Measured outcomes.

An additional outcome: feature prioritisation improved by 80% as a direct result of the structured research programme — 21 user interviews fed directly into the product backlog, ensuring design decisions were evidence-based.

How I led and influenced across functions.

As the individual contributor UX owner on this platform, I was responsible for the full design strategy — not just execution. Key collaboration activities:

• Conducted and synthesised 21 user interviews independently — translating raw research into prioritised design requirements presented to the product team
• Defined information architecture and workflow strategy for the unified platform — presented and aligned with engineering leads before any wireframing began
• Used IBM Bob AI Agent for Figma prototyping and GitHub for design-engineering collaboration — reducing handoff friction and enabling faster review cycles
• Presented design strategy and research findings to IBM product leadership to secure alignment on the platform consolidation approach

What I learned.

• The biggest design challenge wasn't the UIIt was convincing stakeholders that consolidating 4 tools into 1 was worth the migration cost. Design strategy requires as much persuasion as it does craft.
• Involving engineers in every sprint review — not just usability testing — was the single most effective thing I didThey caught edge cases that no amount of desk research would have surfaced.
• If starting again, I would run a service blueprint exercise earlierMapping the full organisational workflow before designing the product workflow would have saved significant rework in the IA phase.