Bot first, agent follow up - The contact strategy of leading Dutch energy supplier, Eneco.
Eneco turned to us in search of an automated, self-serve solution, capable of handling thousands of customer enquiries and transactions.
With an average of 1.5 million calls per year, Eneco aimed to drive traffic away from its flooded contact centre by offering its customers a world-class digital experience with 24/7 support.
“The ability to control our live chat availability helps tremendously when we are experiencing periods of high demand.”
At the end of 2018, Eneco began to develop an Intelligent Assistant solution, powered by our advanced NLP and conversational AI capabilities. Ruud Huigsloot, Digital Service Specialist at Eneco adds - “DigitalCX, now Conversational AI Cloud, was already a part of the Eneco architecture and we were already satisfied with what this conversation platform could do.”
Prior to developing the bot, Eneco decided they would follow a ‘Bot first, agent follow up’ contact strategy. This type of strategy involves offering the customer a warm and seamless transfer to a live chat agent when the question cannot be answered by the bot. According to Ruud, the most important requirement for this type of strategy to be effective is ensuring that the transfer is warm – “warm means the ability to capture the whole conversation the customer had with the bot and transfer that conversation to the agent with the right skills. This reduces average handling time as customers don’t have to repeat themselves to the agent.”
The Digital Assistant went live in March 2019, providing customers with answers to FAQs and navigational support. To help reduce traffic in the call center, the bot was placed in highly visible areas on the website; in the top navigation bar, and on the main customer service page.
Shortly after launch, the Eneco Live Chat feature was integrated into the bot, offering users a warm and seamless transfer when their question(s) can’t be answered by the digital assistant. Additionally, with Conversational AI Cloud, Eneco can control its live chat availability, deciding when and where live chat should be offered.
Fast forward 6 months, the bot now handles 19,000 conversations per week. In addition, the recognition rate – a measure of how well the bot recognizes the customer intent – currently stands at 90%, having risen from 65% shortly after launch. With out-of-the-box dashboards available in Conversational AI Cloud, the recognition rate is expected to rise as the Eneco content team continues to analyze the dashboards & optimize the chatbot on a daily basis. The dialog completion rate – the percentage of users who start a dialog and get to the end of the dialog – currently stands at 68%.
According to Ruud, Eneco has also benefited from integrating its chatbot interaction data into other dashboards. “Besides the extensive DigitalCX reporting dashboards, we also integrated the bot questions into Google Analytics and our existing KPI dashboards. This gives us a very clear overview of what customers are asking on each webpage. This type of data is discussed with the content managers of the website and is used to continuously improve the customer experience. The data we see is very promising.”
Moving forward, Eneco plans to leverage Conversational AI Cloud's ease of use when it comes to building transactional dialogs.
In the upcoming 6 months, Eneco will integrate the chatbot with its CRM systems to automate certain transactions such as taking out an energy plan and reporting a change of address.
In addition, it plans on expanding the bot into other digital channels to cater to a wider audience. “Most importantly, we will continue to optimize on a daily basis to ensure we do better every day” adds Ruud.
“More or less by coincidence, we developed our chatbot during our so called ‘low season’. In the summer months, people rarely contact their energy supplier. As such, we see a significant decrease in calls and web traffic during this period.” Ruud recommends that other companies who are looking to utilize chatbots, take a similar approach. “By developing the solution during your ‘low season’, you have more time to fine tune the solution and to start learning without the enormous pressure of a high volume of calls.”
“It is not a project. For us, it is definitely not a project. It is a new way of dealing with digital customer contact” adds Ruud. “The reason our last attempt failed was because we treated it as a project. We set up a project, we put some effort in it, and said to ourselves it will work. This is all about maintenance. This is all about knowledge. And it’s about the long run. It’s not something you deal with in 2 months. We’re building new fundamentals for digital customer contact in the chat and voice bot era. A very important ingredient in our service strategy.”
According to Ruud, collaboration between the digital team and customer care team is key. “It should be the customer care team who set up and maintain the content in the bot. They’re the experts in customer language and know how to phrase things properly. At Eneco, we started by setting up the chatbot our way (the digital way) and once the customer care team got involved, they told us that we had to rephrase everything.”
With various dashboards available in Conversational AI Cloud containing data in (near) real-time, learning starts almost immediately after launch. Ruud leaves a final remark - “Just start. You learn from day one. What you see when you log in to Conversational AI Cloud is what customers are asking. The benefit from that is huge. We use the chatbot data & feedback to improve our chatbot, but we also use it to actively improve our website content.”
Select a region to show relevant information. This may change the language.
Is this region a better fit for you?