The robots aren’t coming, they’re already here.
“Here” is many places, including call centres, Genpact’s global retail consulting leader Raj Bose said. While some bristle at the continued automation of our lives, ask them about the last few call centre experiences. When dealing with people their opinion is highly dependent on the knowledge of the person, how busy they are, and a host of other factors. Quality assurance staff only monitor a fraction of calls, so improvements are likely slight.
[caption id="attachment_89967" align="alignleft" width="300"]
Why not train your AI systems instead? A properly trained AI system can organize calls into different categories, so the customer gets their needs met more efficiently. Banks can more easily track compliance efforts.
When considering how AI can improve your service delivery, it’s first important to differentiate between the technology’s broad and narrow forms, Mr. Bose cautioned. The narrow is targeted to specific use cases.
“What we’re really looking for is places where it’s repetitive onerous tasks that do require some level of intelligence that can be automated,” Mr. Bose explained.
The ideal situation is where Genpact becomes involved with a client as early as possible, he added. That allows them to clarify AI’s limits while understanding the desired outcomes. Put the two together and they can identify what role AI can play.
“Our conversations are much more business outcome focused,” Mr. Bose said. “We think through it and identify where AI makes sense.”
Compliance matters are much more complex, Mr. Bose said, as not only do companies have to ensure that but also that customers are satisfied and always have a quality experience. The standard check is to record a percentage of calls and listen to them later.
“That’s onerous, expensive and after the fact,” Mr. Bose said. “Using AI and natural language processing we can actually record all calls. The AI tools give us the capability to listen to calls and transcribe them into text. We use natural language processing to determine if the agent said all of the right things to be compliant.”
Genpact is actually able to assign a score to each call so the client can determine if each one meets their numerical standard for an acceptable experience.
Genpact’s ability to quickly help clients improves over time, as their catalogue of rules expands, Mr. Bose said. Each arrangement with a client produces a set of parameters they want to define a successful experience. Those processes are saved so future businesses with similar needs can benefit from Genpact’s earlier efforts. Their list of rules is more than 40,000 long.
While the call centre will look different, people are still involved and they also need training, an aspect which is often overlooked, Mr. Bose said. There often is an initial fear of technology, which can change the types of work the remaining people will do. Some become AI trainers, others monitor data.
“Any technology has an impact on the workforce,” Mr. Bose said. “There aren’t bank tellers anymore as mobile banking grows. It’s a natural evolution.”
Employees also get frustrated when they know there are better methods in the marketplace but their employer doesn’t use them he said.
Before a company goes too far down the technology road they should revisit every stage of every process, Mr. Bose advised. AI aside, are the processes efficient? Are they effective?
“A key element of our assessment is to look at how they are doing to make sure they are at their best before we’re applying the technology,” Mr. Bose said. “It’s silly to apply technology to a bad process.”
Contributors
