Persona to patterns
Unlocked a unified understanding of warehouse managers’ workflows, influencing 3 roadmap priorities.
Background
The company relied on fragmented tribal knowledge and siloed personas, with no shared view of cross-product flows. Insights were undocumented, inconsistent across markets, and rarely shaped product decisions. In logistics, a generic “user” doesn’t work as the workflows shift with culture, infrastructure, and regulations. Without a scalable, evolving system of user understanding, products missed context, causing failed features, low adoption, frustrated clients, and costly rework.
Problem Statement
Users experience friction with complex, multi-step dashboards. To ensure the LLM delivers immediate value at launch, it is necessary to identify and prioritize high-impact pain points and train the model on these critical workflows, enabling it to address user challenges effectively from day one.
Approach & Aim
Hypothesis
Prioritizing the AI copilot’s training on the most used, complex, and time-consuming dashboard flows will improve usability and adoption among target users.
Goal
The research aimed to identify top dashboard flows causing friction, segment users by usage and behavior, validate the need for a conversational assistant, inform AI training priorities and train the LLM.
Scope
In scope: dashboard flow analysis, user segmentation, survey design, training AI copilot on workflows. Out of scope: development of new AI features outside current dashboard flows.
Key Insights
Information-oriented heavy users are primary adopters of the AI assistant; they need step-by-step guidance and analytics support.
Top 10 most used, complex, and time-consuming flows highlight critical user pain points and was faster to train LLM on them
Survey feedback confirms strong interest in a conversational assistant for workflow simplification.
Step-by-step flow mapping significantly improves AI training efficiency and task accuracy.
Impact
Business Outcome
Increased efficiency in operational workflows, reduced errors, and higher adoption of AI-driven interactions for dashboard tasks.
Team Outcome
Data analytics team gained clear priorities for training and improving the AI assistant; engineering had structured flow-based guidance for automation.
Stakeholder Outcome
Alignment on high-impact flows, consensus on primary user segments, improved roadmap clarity for AI integration.