Cox Communications VUI
Cox communications needed a way to improve their customer service experiences. They challenged us to find innovative digital ways to do so.
The Case for Voice
Cox Communications has a business challenge as most in the telecommunication space do; customer service is lacking, the business spends a lot of money on call center costs, and most customers walk away feeling their problems are not resolved quickly - or at all, causing more frustration. Cox challenged us to find alternative digital ways to help customers resolve the top reasons they call for support. As we immersed ourselves in the data we found that many issues could be resolved in a more efficient way. For a variety of reasons, including connecting customer data to the call center simple needs like "What is my login and password" or "Help paying my bill" still bubble up to top reasons people call. Through third party consumer research, we know that the simpler and easier it is for customers to get the information they need autonomously improves the customer experience.
As we began to investigate solution platforms, we found some work for some customer issues, while other help might be needed through traditional means like phone or faq’s. In looking at the top customer issues, we found that “internet connectivity” or “help understanding my bill” might be better served to be handled by bots or simple voice interactions through google home or alexa.
The project became very timely in regard to voice device growth. During the 2017 holidays, Alexa was the top item sold through Amazon and since Google Home started shipping in October 2016 Google reported that they sold more than one Google Home device every second. Other forecasts that helped the business align with my POV was that by 2020 50% of all searches will be done by voice
By 2020 50% of all searches will be done via voice
So, how did it all come together?
Alignment: Affinity & Empathy
First, I gathered a diverse group of disciplines. Those that knew the business, the psychology of people, customer behavior, technology, & those that know how to persuade. I led the charge for our workshop gathering all of the data we had at our disposal. Then, using design thinking methods like empathy mapping and affinity mapping we were able to quickly come to alignment and explore solutions. We identified areas of opportunity based on data we had. Empathy mapping helped us align on proto-personas which were helpful in the process of designing our Skill. We applied them quickly to ground our solution in real-world user needs and began to explore possibilities.
Since we’re designing a dialogue, it would make sense to use a script to model the conversation, right? For the Voice project, I used Google Sheets to write a script that represent an exchange between the device (written in black below) and the user (in blue). Each column represents interactions over time, and items in bold are scenarios like getting help with login and password for example
Once we had a general conversational script path, we needed to collect information from users to understand how users refine a request so I could give them a useful response. Prototyping helped take the scripts from abstract to reality. I built flows out in a software that allowed the back and forth interactions come to life through an actual device. I was also able to make some hypothesese to incorporate additional utterances with the software. This was important to me because I needed to refine the ideas and get something up and running to test with people.
In our first test, I gave the user a strong overview because he would be the one talking to Alexa, but for other tests I gave no briefing at all! For later tests, I created a simple explanation to let other participants know what they were helping us do and how to join in. We found that this helped set expectations, reduce awkwardness, and ease the burden on myself as a moderator. Along the way, we found many helpful insights about the proposed experience that gave us clues into how to solve for different utterances, uncovered additional mental models and really helped us refine the prototype to a state to begin documenting in depth conversational flows and finalize a well defined prototype that the developers are now using to build this exciting skill.