Amazon
Caracara Chat
Caracara Chat is the text-based channel in the next-generation customer service handling platform. Complex customer issues that cannot be resolved though self-service pathways and require human intervention can connect asynchronously with skilled customer service agents (CSA).
Team
Customer Service Tech, Customer Handling Experience
Role + Responsibilities
Lead designer
UX Strategy
Interaction Design
Overview
The Customer Service organization at Amazon is developing a new customer handling platform to replace the current system. As part of this transition, channel expansion to enable increased routable volumes beyond the voice channel was required. I was brought in as part of the contact handling UX team to lead the customer experience effort for adding the text-based channel to increase routable volume. The urgency for this effort was never more evident than during 2020 when Amazon’s global customer handling capabilities were drastically impacted during lockdowns.
Beyond channel expansion, there were numerous customer experience issues that required an overhaul of how customer service agents (CSAs) engage with customers more efficiently to reduce customers repeating themselves when they reach a live agent, shorter wait times, and faster issue resolution.
Research
I worked with our researcher to identify the pain points and gaps for the chat channel. We first looked at the current system by speaking with associates on what worked, what didn’t, and what would be nice to have. Then, the researcher and I ran a design sprint to generate and iterate on explorations we could test with our CSA population.
The key findings that needed to be solved for in the final designs were:
Customers had to repeat intent during each contact, either in self-service or with live agents
Concurrent chat sessions produced significant cognitive loads for our CSAs
Asynchronous nature of chat left the CSA in a state of uncertainty, unsure when or if the customer will respond
Legacy workspace was difficult to manage, increasing chances of missed contacts or errors in resolutions
Vision explorations for overall layout and concurrent chat events.
Prototype
Opportunities + Challenges
The necessity for enabling channel expansion was made clear during the pandemic where many people - including our CSAs from around the world - were forced into lockdown, limiting the capacity to help customers. The chat in the legacy system was never in more demand as increases in orders from homebound citizens also resulted in increased calls to customer service.
While it did the job, we had opportunities to refine the experience as well as include new functionality to better improve the customer and CSA experiences. These include service integrations such as ML-based intent gathering so the CSA knows what the customer is calling about, concurrent chats to minimize customer waiting during peak times, integrated translation services to break down regional barriers, and intelligent assistants for more effective and efficient contact handling.
And we also make continuous accessibility improvements so that we are inclusive to all who use our product.
Contact summary explorations incorporating an early model for intent gathering.
Outcomes
While our minimum lovable product (MLP) is out to a limited number of CSAs, I still can’t share much of the product designs and initial results. Additional outcome details may be shown on a case by case basis.