Conversational AI Software Integration Study
I led UX research to explore how a new Conversational AI Platform, designed to automate personalized conversations across Email, Chat, and SMS, could be effectively integrated into our existing customer support software.
Goals & Objectives
For this research came from a desire to baseline the as-is experience, to support the evaluation of the redesign.
Simultaneously, the research aimed to uncover the initial reactions and potential opportunities for improvement
Overview
Role: UX Researcher
Contributions: I designed and led two phases of user research with testers and evaluators. , conducted interviews, synthesized findings, and delivered recommendations.
Method: We combined usability testing with observation to understand how people navigated the interface, how they interpreted system behaviors, and where they struggled.
Baseline Research: The goal was to measure the existing experience.
Unmoderated testing: with Marketing Managers and Sales Managers
Hard screening criteria:
Role: Sales & marketing managers
Was involved with managing marketing campaigns, among other responsibilities
Min. 500 leads/month
Pure analysis: Due to time pressure, I needed to pivot; this testing was conducted by UX experts. Moderated Workshop & Unmoderated testing with experts.
Key Findings
Visual overload: Users struggled with crowded icon layouts and text-heavy screens that weren’t scannable at a glance.
Broken expectations: Some system behaviors (like auto-renumbering message cards) didn’t align with users’ mental models.
Lack of feedback: Missing guidance left users disoriented in flows, causing confusion and repeated navigation.
Outcomes
Provided actionable insights that directly informed design decisions.
Aligned the product team around user pain points and priorities.
Reflections
Although the software was ultimately not incorporated, and the third planned research phase did not take place, the earlier phases had an immediate impact.
Our findings guided design refinements, aligned stakeholders on user needs, and provided a validated foundation the team could build on if the project were revived.