The Challenge: Inefficient Customer Service for Callers
Ring receives more than 8 million support contacts from their customers (or “Neighbors”, as they’re known) annually. These customer service contacts take an average of 15 minutes of handle time to resolve. With a cost of $1 per minute, this represents a significant expense to serve customers.
In order to reduce those expenses, Ring wanted to automate the authentication part of the customer service process to decrease the time customer service representatives (CSRs) spend on each call, while also deflecting entire calls surrounding order tracking to self service through an IVA. To do this, they needed help creating the call flows, both from a conversational design and technical architecture perspective.
Our Process: Human-Centered Design Agile
Using our HCDAgile methodology, we partnered with Ring’s key stakeholders to understand the as-is state of their current IVR. We documented all of the existing user flows and created a journey map, noting friction points and opportunity areas. This aligned the involved teams (Ring Product, Ring Software, and 1904labs) on specific areas of focus and allowed us to create an MVP approach that prioritized quick wins and ROI while also ensuring that a foundation was created to meet Ring’s longer-term vision.
During the ideation phase, we continued to work with stakeholders and leveraged existing call center data to create a baseline of insights that was used to inform our designs. Architectural wireframes were created that accounted for each step of the users’ experience, as well as any necessary integrations to enterprise systems. Building on the wireframes, our conversational design team worked with stakeholders to build the first iteration of each module.
To monitor performance and optimize the modules, a central repository of data was created using Amazon S3. Working with the stakeholders, it was determined that some means of monitoring the module performance during development and early deployment was important. This allowed us to create a more user-driven experience and “tune” the bots based upon user data. The developer dashboard was created using Amazon QuickSight. Because of the central repository, Ring could also pull data into other existing reporting tools for consumption there.
The Solution: An Intelligent Virtual Agent for Self Service
To complete this first phase of Ring’s long-term vision for customer service, we built an IVA consisting of two modules that will result in significant savings:
- An authentication module that guides callers to authenticate their identity before they reach a CSR. This removes the task from the agent, saving them time, and allows the caller to arrive at the agent having already been authenticated.
- An order status module (called "Where Is My Order" or "WIMO") that allows Neighbors to check the shipping status of their order and receive a text with tracking details.
Both of these modules were developed using Amazon Lex and were integrated into Ring’s existing IVR, which utilizes Amazon Connect. The IVR routes the caller to the appropriate module as needed. In the event that the module cannot enable the caller to self-serve, it is designed to route the caller to an appropriate agent.
The data from the module interactions was captured and ingested in real time and then stored in an Amazon S3 bucket. Then, by executing Amazon Athena interactive queries, we generated a granular view of individual interactions and also created the ability to analyze how the IVA is performing at a macro level. This allows further enhancements of the IVA in the future by determining what it’s doing well, additional phrases it needs to understand, and how to continue to tune and improve it in the future.
The Outcomes: Significant Savings and an Improved Experience
Millions of Projected Savings
Ring receives approximately 4.3 million contacts per year that require CSRs to spend 1-2 minutes authenticating the caller - a process that will no longer be necessary when the caller communicates with the IVA. All of this adds up to nearly 8 million minutes of agent time saved yearly, bringing cost savings of the same number. The WIMO module, meanwhile, is anticipated to handle almost 250 calls daily, saving 12 minutes per call and a projected $1 million per year.
Improved Caller Experience
While the cost savings benefit the business, the new IVA will also make for a smoother, quicker experience for customers. The time saved on authenticating with an agent also means they have less back and forth time and can be helped more efficiently. Since the implementation of the authentication module, Neighbor adoption has increased 25%. Meanwhile, those calling for their order status won’t have to wait in a queue, saving them nearly 9 minutes because of the efficiency of the IVA.
More Self-Service in the Future
The initial self-service framework was built to be extensible, meaning more options can be added in the future, and it can even be used across different channels. Ring has plans to continue this work this year, building on the start we provided.