End to end B2B AI analytics platform design, from UX/UI to a 0→1 design system and data visualization
Client
Causal Foundry
Year
2024
My Role
Lead Product Designer
Team
1 Software engineer
1 Front end dev
1 Data scientist
Timeline
14 weeks
Overview
I led design from 0 to 1 for a B2B analytics platform used by healthcare organizations to send interventions and track its progress. With no existing product or system, I created the design foundations from scratch, including the component library and chart framework, and designed the first core product surfaces with scalability in mind.
My Role
As the lead product designer, I was responsible for defining the product’s UX and UI direction from the ground up, while making sure the information was accessible to end users. I partnered closely with data scientists and engineering to set accurate standards for MVP while establishing patterns for future scaling. I designed the core analytics workflows, chart framework + interactions, component library/tokens, documentation, and design standards.
The Problem
There was no existing platform, no established patterns, and no shared UI language. We were working with complex data that needed to be visualized in an accessible way. We also needed to deliver an MVP quickly while also building a foundation that would scale.
Define an information architecture and interaction model for analytics workflows
Establish a cohesive visual language and reusable components
Create chart standards that would remain consistent across future features
Design for real-world data conditions, not ideal datasets
Enable the team to ship quickly without creating design debt
Target Users
Our primary users were project or operation managers working in healthcare NGOs. They needed to spot trends, compare cohorts, and translate insights into action, often under time pressure and with varying levels of data literacy.
User research also revealed that many of our target users had limited experience with analytics platforms and data visualizations. To support them, I designed simplified task-based workflows for creating interventions, along with charts and graphs grounded in industry best practices, iterated through biweekly demo sessions with our partners.
Intervention workflow wireframes iterated with partners
UI flow showing intervention creation and monitoring
Design system snippet, charts and tables
I designed a scalable chart and visualization framework that balanced usability, visual clarity, and build feasibility.
A central focus of the platform was the data visualization system. I worked with engineers to design a flexible charting framework that scaled across use cases and data types. This included modular chart tiles with responsive layouts and interactions such as tooltips, filters, zoom, and legends, along with solutions for edge cases like missing or minimal data. Our goal was to improve clarity, consistency, and scalability across the platform.
Key Process
Standardized chart anatomy across the product.
I introduced a shared structure for titles, axes, labels, legends, tooltips, annotations, and states so users could interpret charts consistently across views. This reduced mental overhead and made the UI feel more intentional.Designed for real-world data conditions, not ideal datasets.
Working closely with data scientists, I built patterns for missing values, sparse data, outliers, and extreme ranges so charts explained what was happening and offered clearer fallbacks.Balanced visual craft with engineering constraints
Some behaviors were costly to build. I partnered with engineering to prioritize the interactions that mattered most, simplify what didn’t, and align on a framework that could grow over time.Built reusable building blocks instead of redesigning screens
Rather than treating each data chart as a one-off, I focused on reusable components and patterns. This allowed the product to expand without reinventing the same decisions every sprint.
Foundation and specs of a data chart component
Examples of chart types and variants within design system
Documentation of chart types, usage, accessibility and pattern on Notion
Complex use case of line charts and filters
For end users, reliable and clearly presented data built trust, reduced the learning curve, and supported faster, more confident decision-making.
Outcome
Shipped the first version of the analytics platform from scratch, including a reusable chart framework and component library. Standardized across 10 chart types and 5 core product surfaces, with standardized interactions and complete empty/loading/error states. We intentionally deferred advanced customization and niche chart types to keep the MVP focused and maintainable.
Impact
Shipping speed: Enabled the team to deliver new dashboard views in 2-3 days instead of 2-3 weeks by reusing components and patterns.
Consistency: Reduced design and implementation rework by standardizing chart anatomy and interactions across 10 chart types.
User experience: Improved readability and reduced friction when switching metrics and views, backed by fewer support tickets and faster task times.