AI & Automation in the Contact Centre - Using AI to improve Customer Experience and drive efficiencies

In this updated White Paper, FourNet considers how Artificial Intelligence and Automation has developed significantly in the last few years and how the latest technology can be utilised to enhance customer experience and customer journeys. We look at how the technology can improve internal and external processes and drive further efficiencies. And we examine how it can be used to assist employees, both in the workplace and with hybrid working practices.

Post pandemic AI

Developments in all areas of AI, and uptake of the technology, was accelerated by COVID-19. Even before the pandemic, analysts were predicting that conversational AI would become as important for an organisation as its own website. Once coronavirus struck, the functionality offered by AI and automation took three of the top five slots in Tech Target’s Customer Experience IT Priorities surveys.

And when it comes to the effect that AI and chatbots has on customers, the study’s findings are overwhelmingly positive.

However, the more complex and urgent the enquiry and the subject matter, the decision-makers’ survey suggests that most consumers would still like to talk to a human agent.

Artificial intelligence can allow organisations to automate some of the more repetitive tasks in the customer support journey. Chatbots can provide customers with answers to basic questions so that they can serve themselves before seeking the assistance of an agent.

Natural Language IVR can enable customers to self-serve via the phone. RPA can automate the completion of basic tasks such as updating a customer’s address across all systems.

For more complex issues where customers want to engage with an agent, AI can be used to route the enquiry to the agent with the right skills to address the issue thereby increasing first contact resolution.

After many years of comparable costs for phone calls and digital channels, the figures from hundreds of UK contact centres in 2023 suggests that digital channels are becoming considerably cheaper than phone contact.

Every contact centre will have their own metrics for measuring the cost per inbound interaction. But Contact Babel’s latest research on mean average cost per inbound contact shows how quickly an investment in a web/app self-service chatbot or Natural Language IVR can pay off.

Customers have long had the option to vote with their feet (or their keyboard mouse) and switch to a different provider if they don’t benefit from a great experience. Improving the Customer Experience in the post-pandemic world is up in the top three priorities for most C-level Execs.

Customers want to know that you understand and care about them.

Artificial Intelligence can help you to provide that personalised experience by serving agents with the relevant information to provide a great experience.

As well as ensuring that a call goes through to the person with the best knowledge to handle that call, AI can also provide agents with contextual information about a caller. This means that your agents will be able to see what the customer has had issues with before, which products they’re calling about and more. That way, your customer doesn’t have to repeat their story endless times to get a result.

AI powered sentiment analysis can help agents identify when a conversation is moving in the wrong direction and provide guidance on how to get control of the discussion again.

AI powered Workforce Optimisation not only enables organisations to accurately predict demand and schedule resources accordingly but can also supercharge Quality Management processes by enabling the monitoring and scoring of 100% of calls automatically rather than just random samples.

Machine learning can then use this huge pool of data to analyse patterns of agent behaviour and characteristics connected with best outcomes, to develop performance and training programs right down to the individual agent level. Gaps in agent knowledge or capabilities can be identified and addressed based on thousands of calls, rather than relying upon manual evaluations which can only process a handful of calls from each agent.

These capabilities are especially important with some agents regularly working from home.

Beyond improving current performance in the contact centre AI can also use the data gathered from day-to-day conversations to make predictions about what a customer might want or need from your company.

AI can analyse huge volumes of data rapidly and can spot trends and offer predictions or forecasts that we wouldn’t be able to come up with on our own.

Most businesses will offer at least some form of customer self- service, for the most part this is a simple FAQ section on the website or an automated IVR. The main objective for AI in most contact centres consists of projects to increase the capability for customers to self-serve. This not only has the benefit of dramatically reduced costs but also an improved customer experience because of higher real first-contact resolution rates using the customer’s channel of choice.

AI powered chatbots make it simple for organisations to automate customer service interactions and deliver a faster, more personalised experience for customers. Customers can interact with organisations from their preferred device through speech, messaging or visual interfaces and receive an instant, rich and consistent experience. Chatbots connect customers with the answers they need through embedded knowledgebases, from documents, from business applications, and from employees. A chatbot can answer questions and action requests at scale with human levels of understanding and deliver instant answers.

Chatbots, or Virtual Assistants, can deliver immediate responses to customer enquiries 24:7. Utilising AI & Machine Learning an AI powered chatbot can understand conversations no matter how a customer may phrase their enquiry and provides instant responses through automation and integration to live customer service applications.

AI powered conversational chatbots are light years away from the earliest versions of chatbots which could be frustrating and time- consuming for customers. Early chatbots were often only able to respond to very specific input, they were essentially a text-based equivalent of interacting with the old “phone tree” IVR call menus.

Understanding humans isn’t easy for a machine but it’s getting easier by the day, as Microsoft’s Chat GPT4 and Google’s Bard illustrate very clearly.

The subtle and nuanced way humans communicate is a very complex task to recreate artificially, which is why chatbots use several natural language principles including Natural Language Processing (NLP), Natural Language Understanding (NLU) and Natural Language Generation (NLG).

These principles enable chatbots to mimic human conversation. The chatbot can identify the underlying intent behind the text a real person types or speaks, then deliver a response that matches that intent. Chatbots with NLP can now “learn” from past conversations and improve their ability to provide appropriate responses and solutions.

On a simple level, a human interacts with a chatbot. If voice is used, the chatbot first turns the voice data input into text (using Automatic Speech Recognition (ASR) technology). Text only chatbots such as text-based messaging services skip this step.

The chatbot then analyses the text input, considers the best response and delivers that back to the user. The chatbot’s reply output may be delivered in any number of ways such as written text, voice via Text to Speech (TTS) tools, or perhaps by completing a task.

Conversational AI bridges the gap between human and computer language and makes two-way communications more natural, without having to configure specific words or phrases. The intelligence comes from the system learning the many different ways that customers communicate their requests to improve the results and understand more.

The most immediate potential of AI powered chatbots in the contact centre is in handling digital enquiries: webchats tend to take longer than phone calls due to agent multi-tasking, and many email response rates take days.

When the chatbot has low confidence that it has returned the correct result, it is able to escalate the customers query seamlessly to a live chat agent, who then has access to the self-service session history, enabling a greater chance of a successful resolution without repetition.

Contact Babel’s average cost per inbound contact see figure 9 suggests that webchat is now significantly cheaper than voice contact.

But whereas only 5% of webchats had any automation involved in 2015, this has grown to 50% in 2022. There is still considerable room for automation, increasing efficiencies and lowering costs.

The time taken to realise the benefits of implementing an AI powered chatbot can be extremely swift. The stats highlighted below show the monthly impact of implementing a chatbot for one of FourNet’s small housing association customers.

Of those contact centres offering telephony self-service currently, a mean average of 18% of inbound enquiries see figure 9 were handled without requiring an agent, according to Contact Babel’s 2023 research; however, a typical IVR solution can only handle a small number of common enquiries and only works well in cases where the caller has a simple request.

A Natural Language IVR can simply be viewed as a chatbot that customers communicate with through voice, rather than typed text. It needs the capability to handle natural language, to use artificial intelligence to determine intent and provide solutions, and to convert speech to text and back again.

Natural Language IVRs can be used as triage to decide who gets automation and who is directed to an agent based on the profile of the customer, their requirements, circumstances and past outcomes.

As with all forms of customer self-service, an IVR must offer the customer an easy way to opt-out and request to be transferred to a live agent, ensuring all the data that has already been gathered is handed over.

In addition to the standard benefits of enabling customer self-service (24:7 availability, cost reduction, taking pressure off agents etc.) encouraging customers to use Natural Language IVR rather than to press numbers on a keypad generates large amounts of data that can be used to further train AI models.

Alongside voice authentication, AI can also be used to power “phoneprinting” which focuses on preventing fraud. Phoneprinting collects information about the call being made, such as Calling Line Identifier (CLI), location, the type of phone being used, the phone number’s history, levels of voice distortion, etc. These factors are then scored and the score will then determine the security processes and questions that the agent is required to ask the caller, speeding up the process for genuine callers, and focusing the tightest levels of security on potentially fraudulent calls.

Once the system has a voiceprint, the customer gets a better experience when they call customer service. For example, the interactive voice response (IVR) system can recognise and authenticate their identity, which enables the customer to access self-service tasks like checking account balances. If the customer chooses to speak to an agent, the authentication information can be passed to the agent, which saves time and improves the customer experience because they don’t need to repeat any information.

The beauty of AI and Machine Learning powered self-service applications means that they continuously learn and improve based on customer interactions. However, to get started self-service tools need to be trained and fed.

Working with a solution provider with sector specific knowledge can help shortcut a great deal of the front end of this process. FourNet’s sector specific expertise means we can help automate a large volume of simple queries using our sector specific templates that consolidate the most frequently asked questions or address the most common issues by industry sector. Responses to enquiries may include content that rarely changes where pre-trained answers from FAQ’s or workflows will guide customers and resolve their query accurately.

For more complex or tailored queries, the underpinning AI technology can be used to access more complex and dynamic content found in documents, knowledge bases, databases, product manuals, and business applications such as CRM & Service Desk.

Where answers are dynamic and constantly changing, the system can be trained to search through specific content that exists in different formats that are typically challenging for traditional search technology to interpret (PDFs, Excel tables, PowerPoint). With embedded AI, information can be intelligently labelled and indexed within your enterprise document library (headers, footers, content, images, tables) enabling smart discovery of precise answers from within bodies of text.

Where dynamic content is stored within databases, it will search information and content found in business applications such CRM systems, Service Desk, HR systems, databases or industry specific systems. Data can be retrieved to help identify customers for Identity and Verification (ID & V), and look-up content to provide users personalised responses.

AI-assisted service provides agents with information that is tailored and relevant to the customer interaction they are having real-time whether by phone, chat or email. AI assistance is particularly useful for helping inexperienced agents, for remote working scenarios where asking colleagues is more difficult and for more complex interactions which may require multiple systems and databases to be accessed.

AI assisted service offers an opportunity to provide timely and effective support to every agent while the call is happening. AI can provide the agent with suggestions about next best action, pull up relevant information from the knowledge base, make suggestions based on customer history and use sentiment analysis to warn when a conversation is going off track. Real-time sentiment analysis can flag

a supervisor remotely who can break into the conversation or whisper coach. These tools have a positive impact on first-contact resolution as well as customer experience.

AI assistance can help inexperienced agents where stupervisors or experienced colleagues aren’t available, they improve the outcome of the interaction in real- time rather than waiting for post-call reviews.

Using natural language processing in real-time speech analytics can identify when an agent is experiencing stress, becoming emotional, or in need of whispered or direct intervention on a call.

Assisted Service technologies became hugely important during the pandemic where a high proportion of agents were working remotely for the first time. The positive impacts on performance, sickness and attrition for those agents who adapted well to homeworking means that many contact centres have pursued a hybrid work strategy with some agents in the office and some working remotely. This has made AI Assisted Service technologies critical for businesses.

AI enabled contact centres can harness the huge processing power of AI to route queries to the right agent. The technology also helps make more streamlined decisions and use sentiment analysis to predict behaviour and improve outcomes.

In today’s hybrid working environment, technology has become increasingly important to assess how agents are performing when working remotely.

Without the correct tools, supervisors have little direct oversight into an agent’s conversations, and it can be challenging to personalise coaching and understand which behaviours can be improved.

AI for Quality Management is playing a huge role in managing agent performance. RPA is being harnessed in these post COVID times to free agent time and increase operational efficiency.

Enable self-service via chatbots or voicebots;

• Recent projects have seen up to a 43% reduction in the volume of live chat requests that can now be handled by AI virtual assistants.
• Quantify the number of simple queries that could be answered by chatbots or voicebots, identify your contact centre’s cost per inbound contact and model the impact that this switch in types of contact has on your overall cost to serve.

Reduce Average Handle Time (AHT);

• AI powered Assisted Service for Agents can reduce the average handle time per call by suggesting next best action or pulling up relevant information from the knowledge base.  This reduces the need for screen switching and streamlines the call.  We have seen reductions in Average Handle Time of up to 30% by implementing Assisted Service Solutions.

• Robotic Process Automation (RPA) can have a dramatic impact on Average Handle Times.  Using RPA-assisted integrated desktop solutions saves agent time switching between screens during the call and reduces Call Wrap-Up times post call by ensuring data is written back to any relevant databases.

• Average Handle Time for webchat can be reduced to zero agent hours for those queries that can be handled by AI chatbots.

  Increase First Contact Resolution (FCR);

• Intelligent Routing ensures that queries are directed to the right agent to handle the customer’s specific issue.  Making AI powered decisions based on the caller identity, customer journey input and data already gathered via chatbot or IVR can route the contact to the agent best suited to resolve the issue.  We have seen increases in FCR of up to 20% after implementing AI powered Intelligent Routing solutions.

• Assisted Service can have a dramatic impact on FCR rates as well as Average Handle Times, we’ve seen improvements of nearly 20%.

Increased retention rates;

• There are multiple research sources that have proven the link between increased Customer Satisfaction and higher retention rates. According to Invesp, improving customer retention rates by just 5% can increase profits by between 25% and 95%. Using the average Life Time Value (LTV) of a customer you can model the impact that even small percentage increase in retention would have on the bottom line.

• Assisted Service can use AI to suggest relevant cross-sell or upsell options to the agent while on the phone.

A study by Aurora Market Research into 1,000 UK consumers commissioned by Contact Babel in 2023 identified the most important factors to a customer when contacting an organisation

Both businesses and consumers agree that first-contact resolution is the most important single factor impacting upon customer experience when contacting a business. A short wait time for a response is also seen as a vital factor. AI and automation can help on both counts.

The great thing about AI in the age of cloud and as-a-service technology is that you don’t have to take an all or nothing approach to the trend.

In fact, we strongly recommend most organisations make decisions with the long term in mind but start with small projects. By this we mean that for those organisations looking to upgrade their overall contact centre infrastructure, now is the time to look at solutions built with AI at the heart which will provide a framework on which to build for the future. But for organisations undertaking large contact centre system upgrades and those organisations with existing estates, we recommend you start with a small self-contained project for a clearly defined business issue or process e.g. chatbots to replace digital enquiries.

It is important that the boundaries of the project are clearly defined and understood, with relevant baseline metrics captured before the project, and clear and achievable outcomes defined so the success of the project can be clearly measured. A limited, low-risk use case that can be simply implemented can demonstrate a quick win to build confidence in using AI within the contact centre.

While quick wins are essential to get buy- in, it is essential to place the project in the context of a longer-term vision. We recommend building a roadmap of linked businesses cases that layout a long-term vision for the strategic use of AI across all customer facing parts of the organisation.