It's AI match

AI routing makes its mark on workforce management

Intelligent routing offers great benefits for both your customers and your employees. We caught up with Telecats’ Sander Hesselink to discuss the potential of intelligent routing. The two main takeaways from our chat: AI is able to make decisions that contribute towards achieving contact centre KPIs, and AI routing is set to take over part of the WFM process.

Smart routing is the answer to increased complexity in customer interactions. According to Sander Hesselink, in an increasing number of cases, a simple decision tree that guides you towards the right answer by selecting two or three standard choices no longer cuts the mustard. Customers’ questions are becoming more and more complex: they increasingly involve multiple variables that ideally ought to be factored into the routing process. And as for the solution? Well, several outcomes are possible. You might be directed to an employee, to a self-service environment such as an app or a web page, or to a voicebot. But the great thing about AI-based routing is that you can actually take all those variables into account. We met up with Hesselink to discuss the potential of this routing solution. 

Decisions, decisions: focus on employee satisfaction or customer satisfaction? 

Another thing that makes AI routing so appealing is that you can also take your objectives into account when setting up your routing system. Is your goal to achieve the highest possible level of employee satisfaction? Then go for sophisticated skills-based routing and use AI for agent assistance. If efficiency is your priority, you can adjust your routing to ensure faster and more accurate responses. And if your aim is to maximise customer satisfaction, you could focus on providing a personalised service whereby the customer might have to wait a little longer, but will be put through to their usual contact person. You can even program routing in such a way that entry-level employees are not presented with overly complicated questions, but still get to handle a wide range of queries.

Things get complicated, fast

You’ve probably already realised that AI routing can make contact centre management very complicated, very quickly. Hesselink’s advice is therefore to avoid doing that and to keep AI routing simple at the beginning. “This means that you first need to have a clear idea of who is asking the question, what the question is and which team member is the best person to answer it. This is already complicated enough, because you’ll have to think about which customer data you’re actually allowed to use to make these decisions. There are also limits to using customer information for this due to the GDPR.”

“On the solutions side, you have to look at employees’ skills – so product and technical knowledge, possibly soft skills and characteristics”, explains Hesselink. “You could match the level of those skills to customers’ characteristics and the nature of their question. Also, if a customer left a positive review for a previous interaction with a particular employee, you might consider routing the customer to that employee again next time they call.”

“Smart routing isn’t just about making your customer service more personalised, you can also use it to respond to current circumstances”

The art of not interfering

Smart routing isn’t just about making your customer service more personalised (just think of file holders or customer owners); you can also use it to respond to current circumstances. For example, when it’s very busy, you can set up the routing engine to make slightly different decisions than when it’s relatively quiet. Or you can give your customer a choice: if they want to be helped faster, you don’t have to make them wait until their favourite employee is available. 

This is likely to lead to a different kind of decision-making process, says Hesselink: “An advanced AI routing setup may make decisions you might not understand at first. For example, it might not route a caller straight through to an employee who is available. The system may have calculated that making the customer wait until a specific employee is available will yield a better end result.”

If you think of AI routing as being a bit like a black box, with the question coming in at one end and the answer being churned out at the other, you will come up against the issue of explainability: European legislation stipulates that AI decisions must be explainable. 

Can complexity and agility live side by side? 

If AI routing becomes a complex decision-making machine that takes into account a myriad of variables, won’t that seriously affect your agility? One adjustment to your product range or service model can turn everything upside down. “The VGZ case study in particular proves that key changes – for example, new products and brands – are relatively easy to implement if the basics are right. Recognising new intents won’t be the main challenge either, that’s more on the solutions side: who’s available to answer which questions? And don’t forget: if you introduce a new product, you also have to give your employees the necessary training.”

Multiple questions

What should you do when customers ask a question involving several issues, such as “I’ve lost my debit card and need to make an urgent bank transfer, but I am abroad?” Hesselink: “When a voicebot asks people to say their question out loud, they usually limit it to what they consider to be their main question. But machine learning is now also capable of learning how to handle compound questions, i.e., the best way to route common combinations. Although you don’t need to dwell too much on exceptions, you do need to recognise that, when you first start out, fine-tuning the system will to some extent be at the expense of efficiency. But that is just something that needs to be done in order to develop the models that will ultimately increase your efficiency.”

AI routing to take over some of the WFM team’s tasks

When organisations start using Speech Routing, and start using it more intensively, WFM is one of the contact centre roles that is affected. “AI routing models will eventually be perfectly capable of predicting which volumes can be expected for which skills”, says Hesselink confidently. “AI routing will be able to take over the complex puzzle that the WFM process can sometimes be today. You’ll still have to define all your available resources yourself, of course. And the role of a WFM team will shift more towards ensuring that the required resources are actually available.”

Increased integration

Hesselink explains that, to date, Telecats has largely focused on deploying technology before the start of the conversation with the customer. But that focus is gradually shifting to the stage between the question and the answer, and so to related areas such as knowledge management (for example, agent assist: a voicebot that listens in on the fly and actively helps the employee) and WFM, by providing predictive insights into workloads and required resources. As a result, contact centre management is becoming increasingly integrated. 

Text: Erik Bouwer (Ziptone)



Curious to see our solutions in action? Fill in your details and we'll contact you to schedule a demo session.​

Contact us by phone

Get In Touch

How can we help you?