Big Tech Company - Customer Support Chat Platform

Enhancing Agent Efficiency and User Experience in Chat Support

 

Context and Background

As an organisation, our team was responsible for designing end-to-end experiences that interfaced with agents assisting users with problems with their accounts.

Historically, our company incorporated many systems to aid processes. Over time, this became a problem with our leading platform, Agent Connect, which integrates over 18 other platforms, digital providers, and software applications. We often described our problem metaphorically as a messy fridge: there is good stuff inside, but we also need to organise or eliminate many bad things.

This project aimed to revolutionise agent efficiency, availability, and response time in addressing user issues through the multichannel platform (text, email, phone, etc.). It aimed to develop new features that employ machine learning and automation to elevate our service by examining repetitive actions, leveraging historical case data, and harnessing machine learning capabilities within our organisation.

My Role

I was the UX Director, leading a team of one senior product designer and two content designers. My responsibilities encompassed reviewing design solutions, providing constructive feedback, and offering support to tackle emerging challenges.

Problem Identification

To understand the problem comprehensively, we conducted many levels of user research, from user interviews, analysing data analytics, usage heat maps, a heuristics analysis (or expert review), benchmarking, and a competitor review.

User research revealed that prolonged resolution times in current chat experiences were primarily caused by the following:

  • Agents are slow to respond or claim cases.

  • Delays in agents sending their first message lead to early customer drop-offs.

  • Agents occasionally forget to update customers on the status of their cases.

  • Some customers abandon the chat, wasting the agent's time.

  • Agents spent up to 15 minutes checking if customers were still active before closing the chat due to a lack of response.

The Solution

My team and I adopted a multi-faceted approach to address these issues to enhance agent efficiency and chat support experience. We collaborated with our cross-functional partners in product management, engineering, data science, and operations to formulate a greater strategy and develop tactical features to drive improvements. Our components included:

Automated Greetings

Problem: Delays in agents sending their first message, leading to early customer drop-offs.

Solution: We introduced a "quick reply" button to automate the repetitive task of greeting users, allowing agents to focus on higher-priority tasks (Figure 1).

Impact:

  • 24% decrease in Time to First Agent Response.

  • 6% decrease in customer drop-off rate.

  • 7% increase in case resolution rate.

  • 88% of agents found this feature extremely helpful and easy to use (Agent survey).

Automatic Case Closure

Problem: Agents spent up to 15 minutes checking if customers were still active before closing the chat.

Solution: We enabled agents to automatically close unresponsive chats, allowing them to allocate their time more efficiently (Figure 2).

Impact:

  • Auto-closing silent user cases after 15 minutes resulted in the ability to handle potentially 700+ more cases per day.

  • A 13.5% reduction in agent turnaround time.

  • Approximately 72,000 minutes are saved per day.

Notifications

Problem: Lack of notifications for new incoming messages in the agent platform.

Solution: We developed a comprehensive notification system to triage and prioritize messages, aiding agents in focusing their attention.

Impact:

  • Reduced Total Resolution Time (TRT).

  • Improved agent sentiment.

Nudges

Problem: Customers are waiting around 10 minutes for agent responses.

Solution: We introduced a nudging system to prompt agents and encourage timely responses.

Impact:

  • Agents provided prompt service, reducing customer drop-off and Total Resolution Time.

  • Reduced the average agent reply time from 23 minutes to 5 minutes.

Conclusion

These features highlight our user-centred design approach and its positive impact on agent efficiency and user experience in chat support. It exemplifies how thoughtful design solutions can streamline processes, improve user satisfaction, and deliver significant results.