0
Loading
hi@salemtello.studio
Redesigning Google's internal WAN Availability Report from a static, screenshot-based document into a dynamic, interactive data visualization dashboard built with PLX
UX Researcher
& Designer
3 Months
The images about the project are limited given the NDA contract with Google. This image is showing only placeholder data.
The internal WAN Availability Report relied on static screenshots to present network interruptions, outages, and trends — severely limiting analysis capabilities and the users' ability to process the data effectively.
Executives and network engineers across multiple departments who depend on clear, detailed views of WAN performance for strategic decision-making and incident response.
A new interactive data visualization architecture built with Google's PLX library, consolidating five main report categories into a single-screen dashboard with 16 new features for filtering, drill-down, sharing, and data specification.
Users had to navigate through a journey of static screenshots to process information about their network's availability. This format made it impossible to filter, drill down, or compare data points — forcing manual analysis and consuming significant time for what should have been routine tasks.
As my first foray into network infrastructure, I dedicated the initial phase to thorough research — including mastering the PLX library that would define the boundaries of my design.
Gathered context on Google's internal WAN infrastructure to understand the full scope of what the report needed to cover.
Deep dive into network architecture, availability metrics, bad minutes, and how different teams rely on this data.
Mapped the users' pain-points on their journey to process information in the WAN Availability Report.
Audited the 5 main report categories, their data presentation, and identified where the static format failed users.
Learned and mastered the internal PLX library that would be used for development — a crucial phase of the research.
Understanding PLX's capabilities and limitations was essential to define which features were feasible and which were not (e.g., no responsive support).
Led research initiatives to gather context and understand the needs of users across multiple departments.
Conducted stakeholder interviews with executives and engineers to build a complete picture of requirements and expectations.
In-depth interviews with developers and engineers to understand the technical aspects of Google's WAN infrastructure and the current report's role in their workflow.
Mapped the complete user journey through the existing static report, identifying friction points, workarounds, and unmet needs across departments.
Performed extensive simulations with test data in the PLX library to assess visualization and data manipulation possibilities — and document its limitations.
Conducted interviews with users from various departments to obtain a full perspective of their needs, ensuring no critical requirement was overlooked.
Given the diversity of user profiles, I categorized feedback aligned with the report's five main categories. Every single point gathered during the interviews was documented — including what was not possible due to PLX library limitations or strategic objectives.
Historical user feedback and documented pain-points from past report iterations
Analysis of how teams were processing the static report and their workarounds
Fresh insights from multi-department stakeholder interviews led by me
Bad minutes logged every 10 seconds — constantly updating data
Alerts for bad-minutes with prioritization and time-range evolution
No more 28-day limit — ability to set wider analysis windows
Quarterly and weekly trend comparison views
Network agnostic view for comprehensive analysis
General pair availability (independent from B2/B4)
Individual event duration tracking for bad minutes
Drill down to specific region-pairs, B2 sector/B4 neighborhood, metroshard and topology domains vs. aggregate level
I created a comprehensive manual explaining how to reach every single point gathered during the interviews — including clear documentation of what was not possible due to PLX library limitations or strategic objectives. This document served as the bridge between research findings and implementation, ensuring nothing was lost in translation between design and development.
Interface iterations developed in parallel with mastering the PLX library — every feature validated against what the technology could deliver.
Replaced static screenshots with a completely new architecture for presenting network data — interactive, explorable, and significantly more space-efficient.
All five main categories and their subcategories consolidated on a single screen, eliminating the need to navigate through multiple pages of static content.
Added 16 new features enabling users to filter, drill down, share, and specify exactly the data they need — capabilities that didn't exist in the static version.
Every design decision was validated against PLX library capabilities. The tool's limitations (including no responsive support) directly shaped the solution's constraints and possibilities.
Initial interface mockups based on user requirements, constrained by PLX capabilities
Parallel testing of each design element against PLX library to verify feasibility
Built test visualizations to validate data rendering, filtering, and interaction patterns
Refined designs based on PLX testing results and stakeholder feedback
One of the major limitations was that PLX did not support responsive design. The dashboard was built for a fixed desktop viewport — a constraint that shaped every layout decision and interaction pattern.
Learning PLX was a huge part of the process. Understanding what was and wasn't possible allowed me to push the library to its limits while setting realistic expectations with stakeholders about what the final product could achieve.
A new data visualization architecture taking just 14% of the original space while adding 16 new interactive features
The redesign enhanced clarity by 82%, reduced time-consuming tasks by 76%, increased user acceptance by 93%, and took just 14% of the space compared to the previous version — all by implementing a new data visualization architecture with 16 additional features to filter, drill down, share, and specify the needed data.
