It's no secret that the customer service models are changing. “Customer service is switching from voice to digital,” said Katie Molinare, a solution engineer at Salesforce, speaking at a 2018 Dreamforce panel. Technology and customer preference is accelerating this adoption: “Customers are choosing the way they want to interact,” she said.
Chatbots can make a real difference to customers. “Chatbots are key to digital integration,” Molinare said. They provide better engagement for customers and more information and a reduced workload for agents. “We think about our agents spending too much time on quick questions like the status of an order when they could work on more complex questions.”
Chatbots 101
1. Do your due diligence.
The first step should be to evaluate if a chatbot is the right fit for your organization, explained Dennis Thomas, CTO of Neuraflash. You need to evaluate the main cases that are coming in and see if the bot could address them. Thomas suggested joining chatbot communities on Meetup and Facebook to learn about other companies’ experience of rolling bots out. “You want to pick problems that reduce the impact on the organization today,” he said. “For simple questions, a bot can reduce strain on an agent.”
2. Brand your bot.
“You want to brand your bot the same way you brand your customers,” Molinare said. “If it's a surfer company you might want the bot to be a surfer bro.” You can design how the bot greets people, and have it correspond to your company behavior. “It's important to create how the bot will talk to customers,” Molinare said.
—Dennis Thomas, CTO, Neuraflash
3. Give your bot a job.
Molinare said that companies need to think about where the bot will sit in their system. For example, a customer agent bot could provide assistance to customers outside of regular hours. For a lot of small businesses, it doesn't make sense to have an agent 24/7 and the bot could fill that gap. “Start with writing down the five questions people always ask,” Molinare said. These are the most important things for the bot to answer. Next, you need to understand its capabilities. If people ask the bot questions about sales orders, have you provided the bot with access to that data?
4. Help the bot escalate.
When a chatbot can't answer a query there needs to be a system in place for the bot to push the query up the chain. “You want a bot like your concierge as you navigate the center,” said Thomas. “It can answer your basic questions and if it can't, it can send you to the right person.” This escalation is important—the customer feels valued and listened to and their service can be routed appropriately. Make this transition seamless by having the right processes in place. And always be transparent about the service, added Molinare. “We want to let customers know this is a bot. They're not there to replication human interaction, they're there to increase customer satisfaction.”
5. Monitor the response.
To ensure the success of your chatbot, Thomas said, you need to look at how customers respond once it's deployed. Remember, while it might solve a business pain point, it's the customer experience that connotes success. Starting with a clear purpose and making sure the chatbot has a deep understanding of different scenarios is key. “You want to be sure you're setting appropriate expectations,” he said. That way, everybody wins.