Generative AI & Conversational UX
Analytics Assist: Leveraging Generative AI to Retrieve Insights
Directed the 0-to-1 vision for an NLP-driven, prompt-based experience letting HR and Payroll practitioners retrieve insights from complex data by asking natural-language questions.
Analytics Assist — NLP-driven insights experience, ADP Analytics platform
The Challenge
HR teams struggled to derive actionable insights from complex data — defining and articulating the right questions required extensive filtering and specialized expertise.
My Solution
Directed the vision for an NLP-driven, prompt-based experience — letting practitioners ask natural-language questions to retrieve insights and build visualizations.
The Impact
- 4-phase alpha-to-North-Star vision delivered
- 5 competitors benchmarked for strategy
- 0-to-1 product direction for ADP's Gen AI strategy
Research & Discovery
Ideation grounded in client and competitive input
As a vision project, discovery centered on internal ideation and structured feedback loops with ADP's Customer Advisory Board before any pixel was designed.
- Affinity mapping & card sorting (design thinking)
- Customer Advisory Board (CAB) sessions
- Competitive analysis: Visier, DataGPT, Intuit, Google, ChatGPT
- Evaluative research on early concept directions
Key Findings
Trust & Security
Users needed assurance their data stayed confidential and wasn't shared outside ADP.
Effective Communication
Users wanted guidance on phrasing questions to get better responses.
Source Verification
Users needed reassurance that answers were sourced from ADP and properly vetted.
See full discovery detail — additional research themes
Respect for User Settings
Users wanted confirmation that individual settings and permissions were upheld.
Response Optimization
Users appreciated tips for maximizing response quality.
Hybrid Access Model
Users preferred a dedicated Assistant tab plus assistant access from any page.
Design Process
From ideation to North Star vision
Discovery
Defined the MVP; ideated with Product, Data Science & Prompt Engineering.
Concept Design
Explored an immersive "Assistant" tab, tested against user sentiment.
Refinement
Placed the access point on the overview page; explored chat UI formats.
Client Validation
Deployed concept-refinement changes to clients for engagement data.
North Star
Delivered a detailed design built to fully leverage generative AI.
The North Star Design
Built around the "so what," not just the answer
Every response was designed to carry tangible relevance, backed by user history integration and guided interaction flows that reduce ambiguity for practitioners new to AI tools.
- Focused on the "so what" factor in every insight
- Analytics & user history integration
- Guided interaction flows for complex decisions
- Contextual relevance in AI-generated results
Analytics Assist — NLP-driven insights experience, ADP Analytics platform
Outcomes & Learnings
As a nascent technology, AI analytics surfaced real challenges around user trust — compounded by AI's tendency to hallucinate against users' binary "correct or incorrect" expectations. Most practitioners were enthusiastic about AI's potential, but those who engaged directly often hit a frustrating trial-and-error cycle — a key input into the North Star design's emphasis on trust, source verification, and guided interaction.
Additional Use Cases
More case studies
Enhancing Customer Retention Through Analytics Redesign
IA and dashboard redesign of ADP's Analytics & Benchmarking platform.
Creating a Seamless Multi-Platform User Experience
Redesigned Executive Manager Insights for mobile, reaching UX parity with desktop.