About the project
Supply Chain Risk Management (SCRM) Analysts needed to identify hidden vulnerabilities across multi-tier supplier networks, but existing database tools required hours of manual queries to trace entity relationships. Analysts couldn't visualize how risks cascaded through their industrial base or quickly respond to leadership requests for information. Through extensive research with SCRM Analysts and collaboration with Implementation Engineers (IEs), I designed Network Navigator: a graph-based intelligence workflow that surfaces hidden multi-tier vulnerabilities through interactive visualization, enabling SCRM Analysts to investigate supply chain risks in minutes instead of hours.
Role:
Lead Product Designer and Researcher
Tools:
Figma, FigJam, Lucid, V0
Team:
2 Product Managers, 2 Data Scientists, 3 Software Engineers
Timeline:
June 2025 - December 2025
Traditional database querying created barriers to uncovering multi-tier supply chain risks and vulnerabilities.
Design an intuitive graph visualization workflow that enables SCRM analysts to explore multi-hop supply chain relationships and identify hidden risks without manual database queries.
Through research with SCRM Analysts and collaboration with Implementation Engineers, design and ship a graph-based intelligence workflow that transforms supply chain risk analysis. The workflow needed to surface hidden vulnerabilities across multi-tier supplier networks through interactive visualization, enabling analysts to explore entity relationships, identify risk clusters, and complete investigations in minutes instead of hours- eliminating the need for manual database queries and custom data science queries.
Understanding Our Users
We began by conducting interviews with our deployment IEs who work onsite with SCRM Analysts 3-4 days per week. Through these initial conversations and analyses of existing client workflows, we identified two distinct analyst personas with different needs and use cases.
Primary User Groups:
HQ Analysts: These analysts are laser-focused on keeping specific programs on schedule and budget by identifying supplier vulnerabilities before they become disruptions. They leverage network analysis to rapidly answer RFIs (Requests for Information) from directors and find vulnerabilities within their supply chain that could cause disruption.
Program-Led Analysts: Strategic oversight analysts who need to instantly see how a single risky supplier could cascade into program delays and cost overruns across their industrial base. They use nodal analysis to brief leadership with exported risk summaries that quantify systemic risk impact and cross-program exposure.
Uncovering Limitations and Pain Points
Through shadowing sessions with IEs and workflow analysis of both analyst types, we discovered that existing database tools forced both analyst personas into inefficient, fragmented workflows that undermined their core responsibilities.
The pain points revealed during early research were true for both personas:
These findings defined our design requirements: build a graph visualization workflow that could instantly surface multi-tier supplier vulnerabilities for HQ Analysts responding to rapid-fire RFIs, while simultaneously providing Program-Led Analysts the systemic network views needed for leadership risk briefings.
Early Workflow Mapping
Based on these insights, we mapped the core analyst investigation workflow that both personas follow: select an entity of interest → explore network relationships → identify vulnerabilities → share findings.
While HQ Analysts use this flow for rapid RFI responses and Program-Led Analysts apply it to broader systemic analysis, the fundamental interaction pattern remained consistent.
Exploring Interaction Patterns and Iterations
After solidifying the conceptual maps, I began exploring how to translate database relationships into an intuitive graph interface. Early concepts tested different approaches: Should analysts start with a search query or a visual overview? How do we surface risk indicators without overwhelming the network view? What interaction patterns enable seamless traversal across multiple relationship tiers?
Through iterative wireframing, I balanced visual clarity with the density of multi-tier supplier data, ensuring HQ Analysts could quickly drill into specific entities while Program-Led Analysts could zoom out to see systemic patterns.
Testing and Validating with Users
As designs evolved from lo-fi to hi-fi, I conducted validation sessions with Implementation Engineers and SCRM Analysts to test core interaction patterns. These sessions revealed critical refinements needed to support both investigation modes:
Key Feedback:
Robust filtering in context - Analysts needed granular filters accessible directly within the graph view to narrow relationships by risk type, entity attributes, or connection strength without disrupting their investigation flow. This was particularly critical for HQ Analysts responding to time-sensitive RFIs
Contextual expansion controls - "Expand by Risk" and "Expand by Relationship" options needed to be contextual to each entity, allowing analysts to explore different connection types (suppliers, investors, beneficial owners) based on their current investigation focus rather than expanding the entire network indiscriminately
Data table integration was critical - For dense networks, Analysts needed a complementary table view showing detailed entity metrics alongside the visualization. This dual-view approach enabled Program-Led Analysts to identify patterns for leadership briefings while maintaining the graph's spatial context
Adapting the Design Through Development
As we moved from prototype to implementation, two critical challenges emerged that required design adaptation.
Pivoting the Technical Approach
Initial development used Sigma.js as our rendering library, but testing revealed critical gaps: the library didn't support lasso selection, click-and-drag manipulation, or badge components on cluster nodes- all essential for the analyst workflows we'd validated. This discovery led us to pivot to a more flexible rendering library, D3.js that could support these interaction requirements.
Incorporating Accessibility
Given the complexity of graph interactions, I partnered with engineers to design comprehensive keyboard navigation patterns and documented interaction shortcuts that enable analysts using assistive technologies to traverse networks, expand nodes, and access filtering controls without a mouse.
The Solution
Through iterative design, technical pivots, and continuous validation with Implementation Engineers and DOD SCRM Analysts, we launched Network Navigator, a workflow that transformed how DOD supply chain analysts investigate vulnerabilities across defense industrial base suppliers.
Visual Intelligence Through Interactive Graphs
Network Navigator transforms complex supply chain data into an intuitive, explorable network. Analysts can click any entity to view comprehensive summary cards with key risk indicators, financial health metrics, and compliance information- providing immediate context without leaving the graph view.
Contextual Exploration and Risk Analysis
The context menu provides node-specific actions that adapt to the analyst's investigation focus. Analysts can expand entities by relationship type (suppliers, investors, beneficial owners, etc.) or by risk category (FOCI, financial health, compliance, etc.), allowing them to trace multiple tiers of connections and uncover hidden dependencies across their supply chain.
Data Table Integration for Dense Graphs
When graphs contain many nodes, the integrated data table provides a synchronized view of entity details and metrics. Analysts can explore information in tabular format, sort and filter data, and identify patterns that might be harder to see in the visual network—making it easier to extract insights and export findings for reporting.
From Concept to Critical Tool
Network Navigator launched to 5 customer deployments and quickly became an essential tool for SCRM Analysts. The pilot validated the need for visual supply chain intelligence and demonstrated strong user engagement and positive feedback.
Usage
155
Analyses conducted in first 30 days
Conversion
20%
Increase in WAU across pilot deployments
satisfaction
5/5
Positive sentiment from pilot deployments
Internal feedback from Implementation Engineers highlighted the tool's impact: "It is landing with clients. We had demos with 3 deployments in the last two days where it demoed incredibly well and user feedback was incredibly positive. 'This is what I was expecting from Ark.ai when we were first sold it.'"
The team actively responded to user feedback and bugs while shipping version 1.1 updates, demonstrating strong traction and the potential for expanded deployment across additional customer organizations.