The warnings say robots are coming for our jobs, but it’s more accurate to say they—AI-supported automation, that is—are taking over tasks that should be automated anyway.
Taking the rote functions out of a customer service agent’s job is the perfect way to leverage AI, but support roles must evolve parallel with the technology. This is where becoming a knowledge-centered organisation comes into play: Turning agents’ collective people power into collective knowledge-gathering and deployment helps your entire business run, and scale, faster and more efficiently.
We learned a similar lesson from the banking industry after ATMs became popular. The technology was a huge hit, and in parallel, the teller role evolved. Rather than tendering cash, bank tellers became essential for building lasting relationships with customers. They were still on the front lines, but represented the company on more complex issues.
The same is happening for customer service agents. For certain things, customers want and expect automation because it’s simpler and faster. Other issues will still require a personal interaction. Think of every time you breezed through that shipping question on your own via a help center, but needed personalised, impactful help with a human being when your question (or questions) got complicated.
These changes have considerable impact on customer experience, but they have an equally big impact on your agents. Here are some things to keep in mind as you implement new workflows and leverage technology for increased productivity and agent satisfaction as well as a good AI-agent relationship.
Automations aren’t shortcuts
Raising agents’ customer satisfaction scores is one of support leaders’ highest priorities, but inconsistent service remains one of the primary causes of customer frustration. Based on these findings, we know that cutting corners on support teams doesn’t pay off, and that spending support budget smartly is key. In other words, don’t conflate efficiency and bare-bones. The role of customer service manager should not be a race to the bottom.
The challenge in customer service is not pushing agents and their knowledge out the door in favour of bots, because depersonalising this function to save money hurts the relationship between your customers and your company. Smart knowledge bases armed with the technology to support your agents—such as innovations that put help center content at agents’ fingertips in the course of solving tickets—are a key element of continued agent success.
An agile approach to self-service leads to nimble, empowered agents
Technology will only take us so far, which is why updated processes are just as important in supporting agents. Among the Zendesk Benchmark, a data index of 50,000 companies’ customer service practices, we learned that companies with an agile and iterative approach to self-service reported the greatest success. A major component of the agile approach is engaging agents in the content creation process and bridging the gap between content creation and agents’ day-to-day work assisting customers.
AI and machine learning have the greatest potential to act as that bridge. When innovations contribute to keeping rote tasks off their plates, agents can devote more time to knowledge management.
Reframe your KPIs
Changing our definitions of the agent role also impacts how we measure it. As we continue evolving from the outdated idea that support teams are a cost center, new metrics for success will become instrumental in the path forward. A huge shift surrounds two metrics in particular: ticket deflection and time to resolution.
For a lot of organisations, the number of tickets deflected remains a key metric—and it’s definitely something to continue tracking as your help center and knowledge management practices mature. But engagement metrics, such as bounce rate and pageviews, become more important as you consider how your help center can best support customers. Or, consider searches in the help center that didn’t yield a result—as you identify who should be contributing help center knowledge and when, it’s essential to know if customers are coming up empty-handed on searches.
Furthermore, as customers continue moving toward self-service, resolution times may go up instead of down, given the complexity of the issue and the time it takes to provide quality one-on-one service, according to a 2017 article in the Harvard Business Review. One way to manage ticket queues is by empowering customers to close their own requests if an article answers their question, which is possible with Zendesk’s Rapid Resolve feature. As agents then focus their time on more complex tickets, technologies such as the machine-learning supported Answer Bot for Agents help them solve tickets faster and more completely by finding and displaying the most relevant article recommendations from the knowledge base.
Not all metrics are created equal, and one metric could spell success while another suggests room for improvement. That’s why a full picture of the customer and agent experience, on all channels, must be taken into consideration. Just like metrics, AI isn’t one-size-fits-all, either. That means being thoughtful about how it’s adopted and deployed, tying technology to their strategy and remaining proactive about defining success.