Smart or intelligent routing offers great benefits for your customer and for your employees. What should you keep in mind with artificial intelligence-based routing?

 

TechUpdate is Ziptone's section on emerging technology relevant to customer engagement. We briefly explain what it is, how it works, why it is relevant, what the pitfalls are and we ask the reader's opinion: Hot or Not? 

What is it

Intelligent routing is a method for connecting a customer seeking contact with the most suitable employee based on artificial intelligence. When determining the best match, the characteristics of both customer and employee are taken into account. Smart routing is one of the ways to service customers more personalized to make. Other names and forms are AI routing, smart or intelligent routing or skills-based routing, but the latter is best known in a simple form without the use of AI.

How it works

In the most optimal situation, both the customer and the customer's intent are identified at the start of a call. The first can be done by entering data (such as a customer number) or recognizing the telephone number. Another possibility is to use an in-app call, which recognizes the customer because he has logged into a my environment. The second, the recognition of the customer's intent, is possible with an IVR or, even better, with routing via query speech recognition. Another option is as a follow-up to a chatbot dialogue where the customer's question is escalated by the chatbot to an agent. Routing based on (open-ended) speech recognition is also known as speech routing.

After recognizing the customer and intent, the AI ​​engine searches, based on current and historical customer data and the customer's intent, for the employee who is immediately available (or, in the AI ​​engine's estimation, soon) and who has the best skills and personal characteristics. best matches the intent and/or characteristics of the customer. With current and historical customer data you should think of age, gender, language, preferences, situation, purchasing behavior, contact history and recent or real-time information; with intent you can also take the context into account in addition to the substantive question, so that you can (also) give priority to certain conversations.

When matching customer and employee, the engine can be configured in such a way that KPIs are achieved. Think of a higher NPS as a result of, for example, an optimal emotional click between caller and employee, a higher first time fix or a shorter average call duration.

With intelligent routing you can better speak of a waiting field than a queue: the queue is no longer built on the basis of waiting time alone, but also on the basis of all kinds of other characteristics.

Why is AI routing relevant

AI routing can contribute to higher customer satisfaction and efficiency. With AI-routing you can align your routing with the most important KPIs for your organization.

Not only can AI routing contribute to a good conversation – for example, employee and customer understand each other better or already know each other from previous conversations – but a question can also be answered faster because the employee is the right expert.

AI routing also offers the ability to prioritize calls. For example, based on urgency (emergency or current problem), or based on customer characteristics such as prospect or new customer.

Not unimportant: with smart routing you can let your employees have those conversations that they are good at. These are often the conversations that employees enjoy having the most. In theory, this can be combined well with self-organizing teams and/or a high degree of autonomy. You can also provide the employee with useful information about the customer, subject and possible conversation suggestions at the start of the conversation.

And finally: with smart routing you can get rid of the IVR. This is useful for rough routing work, but not exactly customer-friendly.

Just pay attention

1. AI routing goes a step further than skills-based routing, where employees are offered conversations that match their knowledge and skills in terms of question or subject. With skills-based routing, it is not necessary to take into account the characteristics of the customer itself. For the simplest form of skills-based routing, the use of an IVR is sufficient.

2. AI routing requires that you know exactly what information you want to route. And that this information is available and accessible in the form of sufficient and reliable data. You will therefore need to have good employee profiles – and keep them up to date – and sufficient historical ACD data. In other words, requirements are set for data quality. Certain properties that you would like to include may not be easy to contain in data, may not be present (such as the customer's language) or cannot be made accessible (for example, data from a logistics chain). Finally, the question is to what extent real-time data is important in serving your customers. Once the AI ​​engine has made a match, reality can change in the meantime – think of aviation.

3. AI routing is more difficult to apply to employees who still need to be trained. For this you will have to compile a specific employee profile and you will also have to define which conversation topics you want to release on this profile. In addition, you will have to develop a clear growth path towards certain employee profiles. The system becomes unreliable if your employee file is not properly classified.

4. This way of routing works especially well for larger customer contact operations, because you will impose restrictions on employee productivity. They mainly get conversations that suit them well. To prevent them from ending up in 'idle' (waiting for a new conversation), you can set thresholds – think of a maximum waiting time for customers waiting for the ideal agent or a maximum waiting time for employees waiting for the most suitable customer. After exceeding the threshold values, you can choose alternative routes.

5. This way of routing only works well with unambiguous customer questions. It is not always clear in advance whether a customer wants to combine several questions or has complex problems. With compound problems (for example, the combination of failing technology and payment arrears), you should think about transferring (or not) to the right employee when the other topic comes up.

6. You must actively monitor the use of smart routing to prevent blind spots in the service and because both internal and external circumstances change. When making adjustments to products or other business processes, you will have to check whether the AI ​​routing needs to be adjusted.

Hot or Not?

Featured, Knowledge base, Technology, tech update concentric, Concentrix Catalyst, Telecats