LinkedIn Marketing Solutions (LMS)- Smart Conversations

Role: UX Researcher

Study type: Evaluative/Usability | Unmoderated

Timeline: August - September 2023

Methodology:

• Research platform- UserZoom

• Recruitment- LMS bulk recruit (6 participants, 1 soft launch/5 full launch), screened in participants with an LMS help ticket request within 90 days of the study launch were ramped into the Beta experience on the live site by an Engineering partner

• Live help assistant experience tested here --> https://www.linkedin.com/help/lms

Problem:

After the LMS team conducted rounds of internal testing, leveraging past support transcripts to validate and score GPT responses, this usability study sought to surface customer insights on the usability of a virtual help assistant with GPT being powered with GAI, to inform the best approach of scaling to a larger user base. The goal was providing a faster yet accurate resolution on LMS support requests, reducing wait times for users, and reducing the need for a live representative to accurately resolve support requests.

Research Impact:

My research led to surfacing the following opportunity and recommendation that stakeholders implemented post study:

Due to most participants experiencing cognitive load because of the lengthy help assistant replies and slow GAI response time, I recommended the LMS team partner with Engineering to improve ease of use and GAI output.

This led to the LMS team testing a behavior called "streaming" to minimize cognitive load on the user’s chat reply experience, which led to perceived timeliness, accuracy, and digestible responses to their help assistant inquiries. This feature is currently in pilot phase.