AI In Customer Service

Artificial intelligence (AI) is set to play an ever-greater role in customer service. As the technology continues to advance, its potential to transform customer service delivery is becoming clearer. Although many consider these to be the early days of AI adoption, generative AI is beginning to showcase its potential across a significant range of applications and is set to play a role in the delivery of simple customer service for businesses. AI-enabled chatbots, virtual assistants, knowledge bases, sentiment analysis and smart call routing all have the potential to support customer service operations alongside skilled and trained human agents.

AI has been able to reduce costs and improve efficiency for businesses operating across a wide range of sectors. However, it is at its most effective in the area of customer service when its use is carefully planned and controlled. It is critical that AI is organised to support the delivery of world-class customer service and deliver efficiency gains in a given customer service application otherwise its use may well be to the detriment of service delivery. We outline how AI can be used intelligently to enhance customer service, its key limitations and our view as to how its use may develop in the future.

How AI Can Enhance Customer Service

AI can enhance the delivery of world-class customer service when it is used in a planned, controlled and intelligent way. The capabilities offered by AI technology, including generative AI, can reduce costs and improve efficiency within both an in-house and outsourced customer service call centre. It can augment the capabilities of agents and even deliver simple customer service in some specific cases.

Delivering world-class customer service is a significant challenge, especially as a business scales and its customer service requirements follow suit. The introduction of AI into processes and wider customer service operations can improve the quality of service that a business provides. It is possible to enhance customer service using AI without undermining customer service quality with the help of a professional customer service outsourcing partner like Impact. AI excels at simple customer service now and is starting to develop further capabilities.

Chatbots & Virtual Assistants

Chatbots and virtual assistants that are powered by AI bring the opportunity to reduce costs and increase efficiency and this is a well documented and significant benefit that cannot be overlooked. They are also adept at helping organisations with scaling customer service requirements to manage seasonal peaks and deal with large volumes of customer interactions quickly and efficiently. Many chatbots and virtual assistants have progressed far beyond pre-determined responses and set workflows and now harness generative AI to dynamically respond to customer requests with conversational responses.

Simple customer service requests such as order/delivery updates, customer policy queries or questions can often be handled by a chatbot or virtual assistant as AI excels at data retrieval related tasks. There is even potential to introduce human-like speech responses to simple requests with generative AI to deliver a more natural and authentic customer service experience. Customers are able to enjoy interacting with a chatbot or virtual assistant in a natural way without having to adapt their own language to suit the technology. When simple customer service requests are managed in this way, human agents can focus on handling complex  queries that will require a skilled, trained and knowledgeable agent to bring about a suitable and timely resolution to the customer service query or issue.

As long as AI-enabled chatbots and virtual assistants are deployed in a planned and careful way, they can deliver exceptional customer service. However, like all forms of technology, they have limitations and should not be solely relied upon to manage every customer service interaction. A significant proportion of these, and particularly those of a more complex nature, are best managed by skilled customer service agents.

FAQs & Knowledge Bases

Online FAQs and knowledge bases have been used to assist customers by providing self-service customer support for many years now, with the early versions consisting of static web pages that were limited to providing answers to pre-determined questions. The development of both AI and machine learning technology has significantly extended the capabilities of knowledge base systems by introducing intelligent search functionality, which makes it easier for customers to navigate a knowledge base and find more suitable answers to their questions. AI-enabled knowledge bases are also able to deliver an extensive level of personalisation, which includes making tailored recommendations based on their previous interactions and user preferences.

The combination of AI-powered search functionality and advanced personalisation has significantly improved the user experience delivered by knowledge bases which sometimes also incorporate AI-powered chatbots to help customers navigate and interact with ease. This has made knowledge bases a helpful tool for customers who are looking for answers to a given query. Introducing a self-service channel like a knowledge base has the potential to significantly reduce costs and increase overall efficiency of customer service operations. They are particularly effective at helping customers resolve simple issues themselves while allowing customer service agents more time to focus on complex issues and requests. When knowledge bases are deployed in addition to skilled and trained agents, they have great potential to enhance the service offering. They can also provide a limited level of out-of-hours support to customers.

Many AI-powered knowledge bases also offer advanced analytics, which can provide useful insights such as customer trends, common issues and customer service interaction times. However, it is important to avoid becoming overly reliant upon these, especially without an adequately resourced customer service team who can resolve complex queries and provide an alternative contact option for those that cannot or do not wish to use an AI-powered knowledge base.  At Impact, we deliver world-class outsourced customer service for clients using all available technology and this is designed to delight our clients’ customers regardless of the channel that they use to interact with us. Why not discuss your customer service outsourcing requirements with one of our specialists by completing our contact form or calling 01794 230 230?

Speech Recognition

Speech recognition technology itself is not a new addition to customer service, but AI-enabled speech recognition with advanced sentiment analysis is now demonstrating extraordinary potential to transform call quality monitoring. When speech recognition technology is combined with sentiment analysis technology, it helps to monitor voice-based customer service interactions with ease, even at scale and across a significant number of customer service agents.

Prior to this technology, customer service managers and call centre coaches would need to listen to calls to evaluate performance, identify improvements and customer service trends. AI-enabled speech recognition technology is now being used to accurately identify features from call recordings, analyse calls and present useful insights. This is particularly useful for maintaining the quality of customer service interactions and driving on-going service improvement. By scanning customer service calls and understanding the voice of the customer, it is possible to make personalised improvement suggestions to call centre agents and also communicate key trends to other relevant departments.

The speech recognition and sentiment analysis technology does however struggle to interpret tonation but this is developing at pace. Although AI technology is currently able to identify positive or negative keywords, its lack of capability in regard to tonation can result in incorrect analysis and drawing inaccurate conclusions from customer service interactions. It is important, therefore, to combine the AI-enabled technology with a level of manual reviews to confirm the accuracy of the analysis. AI based call analysis is very successful at identifying which calls should be listened to first.

Smart Routing

Smart routing of customer service interactions based on the nature of a query/issue to relevant departments such as returns, complaints or technical support, for example can drive significant query handling efficiency gains. With the language model that has now been built by the AI tech industry in general, the ability to ask a customer what they are calling about and make apparent sense of the response is a stunning demonstration of how the technology is evolving. Gone are the days of repeating “in a few words, what can we help you with” which has been the source of frustration to callers for some years now, the introduction of AI capabilities to existing smart routing technology is now able to improve the accuracy and categorisation of queries received across a range of channels.

AI-powered smart routing technology is able to identify words from an interaction and automatically suggest a suitable routing option based on this. The technology is able to complete smart routing on a multi-channel basis across phone calls, emails and e-forms. By routing queries to the most suitable department, customers are able to enjoy a faster query resolution while customer service agents are better able to prioritise queries and deal with issues that are suited to their areas of expertise.

Key Limitations Of AI In Customer Service

While AI in customer service has significant potential to reduce costs and improve efficiency, it is subject to a number of key limitations. The extent to which you adopt AI technology is likely to be influenced by a range of factors including many of the limitations that we discuss. Whether you are concerned about it restricting access to human customer service agents, unconvinced of its abilities to handle emotive interactions or concerned about its capability for complex customer service, they are all important factors to consider.

It is often the case that a blended approach to customer service delivery that draws on the best capabilities of AI while retaining the specialist expertise offered by human agents delivers the best results.

AI Can Undermine Quality By Restricting Access To Human Customer Service Agents

In some cases, when AI is over-adopted in customer service, it can restrict access to human agents. While this can deliver cost savings and efficiency gains, it can be to the detriment of the customer experience as not all issues can be resolved by AI. If for example, a customer starts an interaction with a chatbot or virtual assistant which cannot later decide to transfer away from AI to a human, the customer will be left frustrated, and it may result in a complaint and even customer loss.

Unfortunately, it is all too common for customer service experiences to be front ended by a chatbot or AI-driven support system and for there to be no option to transfer to a human if the AI customer service assistant is unable to help or resolve an issue. Making it easy for customers to switch to a human led customer service experience during an AI-powered interaction is key to maintaining world-class customer service experiences.

In some cases, businesses that adopt AI in customer service will resort to hiding their phone number or email address, to reduce costs and force customers towards AI, which is frequently to the detriment of customer service. In order to retain and improve customer service quality, it is critical to provide customers with a choice as to whether they use AI or interact with a human customer service agent. This is particularly important when it comes to complex queries where AI is unable to assist.

When AI is used to help with relatively simple requests that are received in high volumes, it has great potential to improve customer service. However, as soon as it is over relied upon, and it frustratingly restricts access to human customer service agents, it can undermine the very customer service experience that it intends to enhance.

AI Is Not Adept At Emotive Interactions

AI is able to respond and manage simple customer service interactions. However, when it comes to emotive interactions where a customer’s emotions are running high, it is far less able to respond in an appropriate way that takes account of a customer’s current feelings or frustrations. For example, if a customer contacts customer service about cancelling an order for a relative placed that has sadly passed away, this needs to be handled sensitively. In this case, the customer will expect a customer service agent that can empathise with them and be understanding of their situation. AI might not be able to respond appropriately, whereas a skilled and trained human customer service agent is much more likely to respond in a way that takes account of the emotive nature of the interaction.

Similarly, if a customer contacts customer service and is frustrated about an on-going unresolved complaint and varies their tone in their response, this might not be accurately detected by AI, which could result in an unsuitable response to their query being given. In this case, a carefully considered response that takes account of the highly emotive nature of the interaction will be required. A skilled and trained human customer service agent is much more likely to be able to respond to the customer and bring the complaint to a satisfactory resolution. The customer will be able to avoid frustrating loop-like interactions and secure a faster resolution to their previously unresolved complaint.

AI Has Limited Capability For Complex Customer Service

While AI is able to manage everyday straightforward customer service interactions, it has limited capability for complex customer service. Customer service operations routinely involve responding to complex multi-layered issues, some of which may require deviation from established policies or charters in specific circumstances. Being able to judge whether a given issue warrants such deviation requires input from a skilled agent, who can exercise judgment and make a call without overly relying on pre-determined rules.

For example, if a customer contacts customer service with a support query about a product that has malfunctioned and is just a few days outside of the 1-year warranty period, an AI chatbot/virtual assistant may apply a pre-determined warranty rule and determine that the customer cannot receive a replacement product. Whereas a human customer service agent may look at issue and take a broader view, which takes account of the customer’s loyalty to the brand and their history of repeat purchases and also the fact that the product in question has a rate of return that exceeds the baseline. In this specific case, the human agent may decide that an exception may be appropriate in this case and either escalate it for further consideration or authorise the replacement themselves if they are able to do so. This may well be in the best interest of not only the customer but also the business itself as they are able to delight the customer, secure on-going loyalty and gain further repeat purchases, thus increasing lifetime customer value (LTV).

At Impact, we excel at complex customer service and provide customer service outsourcing solutions to some of the UK’s leading brands. Our skilled and adaptable agents go above and beyond to resolve even the most challenging customer issues. We pride ourselves on being able to deal with complicated customer service requirements at scale for our clients.

Relative Infancy Of AI Technology

AI technology has been available in certain forms for several years, which makes it all too easy to assume that it is an established form of technology that is tried and tested for a range of business applications. However, its application to customer service is relatively new and it is far from being a well-established and tried and tested technology. While there is an imperative to innovate and adopt new technologies such as AI to drive cost savings and efficiency gains, this must be carefully planned and managed. Adopting AI at scale with limited regard for its efficacy for certain customer service applications has the potential to cause significant disruption to customer service operations, particularly for organisations that operate at scale. It can even undermine the customer service quality and cause further customer service issues that can be costly to resolve.

Although it is clear that AI is very much here to stay and that it will play an important role in customer service, it does need to be carefully adopted, managed and viewed as a complimentary tool for customer service operations rather than something that can mostly replace skilled and trained customer service agents. Human-led customer service that delivers a personal touch with empathy to customers is not going to go away. Balancing empathy with efficiency is key to providing world-class service that delights our client’s customers and delivers beneficial outcomes.

Security, Privacy & Ethical Considerations

The security and privacy of AI platforms can be a concern for many businesses, especially as adopting AI usually involves exchanging information that can sometimes be sensitive and private with AI systems. Ensuring that every AI-enabled system that is used maintains robust security standards and has adequate security procedures in place to protect data is crucial. Where AI systems record information and use it to improve the large language models (LLMs) that power many of them, there are potential security and privacy implications for businesses to consider. While there is a balance between utilising AI and maintaining security & privacy standards, it is a careful balance to reach and maintain. For most businesses, maintaining security standards is a legal requirement and crucial to maintaining the trust of their customers.

In terms of ethical considerations, AI in customer service raises a number of issues. The nature of AI technology, including generative AI, introduces the potential for significant errors and bias in addition to the general limitations that it is subject to. As AI generates answers to queries and questions based on patterns, it can often miss the context in a customer service interaction. Unfortunately, this can result in it generating incorrect answers at best and potentially offensive and inappropriate answers at worst. These can in themselves raise further ethical and potential legal issues.

There have been cases where inappropriate responses that are lacking in empathy have been generated by AI and sent to customers who are facing challenging circumstances including bereavement. This underlines some of the limitations of AI and suggests its use in customer service must be carefully planned, controlled and monitored.