Customer Service Automation
Are you looking for ways to increase the degree of automation in your Customer Service? Would you like to know more about the technologies that can improve your Customer Experience and Operational Efficiency? If so, this blog post is meant for you.
Matti Toivonen, our CXO, was a guest speaker in the Ask Me Anything session organized by our partner Digital Workforce. During the webinar, Matti answered questions around Customer Service Automation, Chatbots, and Conversational AI.
Here are the answers to the most burning questions:
Why should customer service be automated?
- Customer needs are constantly changing. The on-going digital revolution has impacted customer behavior. Customers have become more demanding on the services and especially the digital appearances of the companies they do business with. This happened already before Covid, but Covid fastened it. Also, we all use instant messaging channels and they have become almost primary communication methods. Most likely you are in touch with your family and friends through some kind of app. The rise of instant messaging has changed the way we communicate in general.
- Customers want to be in charge of choosing the channel they want to be in contact, and we believe that this channel is a digital one. People also expect to get service when it is suitable for them, so a lot of service transactions happen outside the normal office hours. People don’t want to queue anything, so they expect to get service fast – and not just any service but good quality service. That puts the service provider in quite a lot of pressure. It is not easy to offer 24/7 services and to organize that kind of operations. So, one should think about what technology could offer as an alternative to those situations. Basically, it becomes more and more difficult to tackle customer demand so that makes room for automation.
- Especially external customer service questions are pretty routine. 50–80% of the questions are standard questions in which users could find the answers to themselves. In principle, all situations where human communication does not add value should be automated. There are cases where you need to show empathy or cases so tricky that they require a lot of investigation. Those are cases where it is worthwhile for a person to handle transactions. But routines and standard questions should be the starting point of automation.
What are the typical use cases of automation?
- The best return on investment comes when there is a reasonable amount of contacts. At the end of the day automation should save time and effort and that happens when we a are talking about a reasonable amount of volume. It’s hard to give an exact number of the amount because it depends on the case but typically if the organization already has an organized customer service team or unit, then most likely there are enough tasks to get started with at least some sort of automation.
- Chatbots are used both in external and internal use cases. Most of the applications so far are external, but the utilization of AI in internal functions is growing. For example, HR and IT have a lot of potential. Example industries vary a lot in terms of the volume of the contacts and the customer base, but typical examples are banking, insurance, and ecommerce. Chatbots can be used also for marketing, for example lead generation and even recruitment.
How does the automation impact customer satisfaction?
- In general, it seems to be hard to get good satisfaction reviews in customer service. And it’s good to remember that people give negative feedback more easily to a computer than to a human. But what we’ve seen so far in the implementation with the clients is that if the chatbot covers most of the topics that the transactions are about, there is typically a good level of customer satisfaction. From a customer point-of-view, if you can get better service in terms of availability then that itself is a positive thing. And if the problem is solved that makes the customer satisfied.
- Automation also impacts employee satisfaction. When the routine work has been removed from people, there is more time to focus on cases that add value. That makes the work itself more meaningful. Automation solutions also provide new roles for people. What we have seen with our clients is that the ones who work with the AI and are in charge of teaching and training the AI have been really satisfied with the work and can sort of multiply their expertise into the system.
What is conversational AI capable of?
- We work with a platform that has natural learning understanding capabilities and is built in deep learning. That itself is a differentiation factor to what type of a bot you’re looking for: are you looking for something that has real AI skills and natural language processing or are you looking for something simpler. That puts the technology quite into a different light. There are plenty of options and one is no better than the other – they just suit in different types of purposes. So, it’s important to think what the role of the chatbot is and what you want to achieve with the solution. We work especially in Nordic languages and our AI understands different dialects. One example is our OuluBot. Oulu is a city in the Northern part of Finland and has its own strong dialect and we have been able to train the AI to understand the Oulu dialect.
- One of the biggest differences between the AI platforms is the easiness of using itself: are the platforms made for AI professionals and programmers or are they low-code so one can manage it with some sort of programming skills or are they no-code platform systems that everyone can manage themselves. We operate with a no-code platform that everyone can learn to master without any programming skills.
- Text, images, videos, files, and location can all be used with chatbots. Voice is coming too. So basically, whatever we see on the web browser is doable also with the bot. The customer can be identified too. And in most cases, it should be. Not knowing with whom we interact the level of service is rather low and general. For example, bank’s customers need to be identified: the bot can’t help you much if it doesn’t know who you are and if you have an account in that specific bank.
How to successfully implement?
- Based on our experience we would say that very often chatbots are seen as an IT implementation of a project or as a programming project which they shouldn’t be. The value is not only in the technology itself but implementing it into the business needs. Then there is a better chance to deliver something valuable and make sure the focus is not only on the technology because the partnership takes care that the solution is right for the usage. One should also understand the actual problem and the root cause behind it in order to implement a comprehensive solution and not just a chatbot. So, to do the automation correctly one should always think about the channel strategy as well. That’s crucial for going forward after the implementation of the bot.
- What we strongly advice in the implementation process is to use the existing data to understand what should be automated – so seeking the truth behind the need. But basically, what our customers say is that just implement fast and good enough, push it in the production, follow up, and do the necessary changes. And that’s it. Those are one of the key steps to manage the quality of the project.