AI

How Bots Automate Customer Service while Lowering Costs

AI bots are not new to automating customer service operations but they have evolved to be more than just basic chatbots that handle a few basic FAQs before handing off to a human agent.

In these unprecedented times during the COVID-19 crisis, AI bots are rising to new challenges and playing a critical role in the automation of customer service interactions. With contact centers and other business operations forced to close and many workers directed to work from home, our virtual worlds are taking on a new meaning. Today, if you call a customer service line or interact on the web via live chat, the likelihood is that the human on the other end is working from a quarantine base. That’s if your call or request is even answered!

AI Customer Service Automation versus Traditional Customer Engagement Channels (Phone, Email, Contact Forms, and Live Chat)

Customer service interactions have been predominantly conducted via phone, email, contact forms, and live chat with an estimated 60% to 80% of customer calls originating from phone numbers found online — through search, maps, Contact Us pages, etc. But all these channels have proven to be costly to service providers as they are dependent on humans and the associated labor costs – this makes it difficult to scale customer support to cover peaks. They are also channels that pose barriers to automation. Implementing AI bots or chatbots is a great first step for customer service transformation.

Inbound calls to a contact center are handled by customer service agents where average handling time drives the average cost per call. Depending on salary levels, country of operation, and the type of queries being handled, the cost of handling customers by phone is high and can be especially prohibitive when it comes to agents responding to routine queries. Also, because the phone relies on voice communication, it is harder to automate responses. Even though agents have scripts and knowledge bases to search for information, it can be challenging to resolve a customer issue efficiently.

For this reason, businesses started to adopt call deflection strategies to try and drive customers to email or live chat channels. However, these are also powered by humans.  Email communication has the advantage of supporting asynchronous communication between a customer and an agent, so makes agent capacity planning easier but it is also a very manual intensive channel. Often it can take several email interactions to resolve a customer issue, potentially spreading the communication out over days. This can incur delays in resolving issues, resulting in poor customer satisfaction or CSAT. 

Although the promises of live chat predicted that an agent would be able to handle multiple chats concurrently, evidence shows that concurrency is often lower than 2 i.e. the average agent handles less than two customers at any one time. Live chat also relies on continuous sessions as an agent deals with a customer until their issue is resolved, or sometimes until a session times out because the customer had to drop the chat. In live chat, the dropout rate can be as high as 25%, compared with 1% on a messaging channel. Similar to phone conversations, the live chat agent needs fast and easy access to different knowledge bases or scripts so it is still quite a labor-intensive engagement channel.

Even though modern contact center technology has helped agents find information and next-best-action recommendations more easily, these traditional channels have still been limited when it comes to automation opportunities. Now with advancements in artificial intelligence (AI) technologies, customer service interactions can become more automated and cost-efficient with the introduction of an AI virtual assistant service or a conversational AI bot.

An intelligent virtual agent is powered by AI software that simulates a conversation with a customer, using natural language. This can be either using voice or text. So whether a customer is asking Alexa to provide the status of an order or is texting in a messaging channel to request billing information, they are interacting with an AI virtual assistant. Sometimes there is confusion between bots and live chat since both are often activated from a business’s website and involve text-based chat. However, there is a big difference in that the former is powered by AI automation and the latter by human labor.  From the consumer’s perspective, it can be hard to differentiate whether they are interacting with a human or with a virtual agent. And the better virtual assistants have become the harder it is to make this distinction.

IWe also have a post on the differences between chatbots and virtual assistants.

Four Reasons for Customer Service to Deploy AI Automation: Access, Deflection, Automation and Collaboration

The results of the pandemic forced many businesses to urgently rethink their customer service strategy and operations. Handling increased inbound requests from customers while trying to manage a workforce that has suddenly shifted to a remote working model is no mean feat. With an increasing need for operational efficiency, customer experience and contact center managers are quickly realizing the multiple benefits of using AI assistants. 

1. 1. Access to Customer Service 24/7: Bots can work 24×7 at no additional cost. This means they are available as much as the business wants and needs them to respond to out-of-hours requests. And if they cannot fully resolve a customer issue out of hours they can take all the details, raise a ticket, and/or schedule an agent to respond when the business reopens.  By extending customer service hours, the contact center can alleviate some of the demand peaks while also providing more convenience for customers. During Covid, helplines and customer support centers became overwhelmed with customer concerns associated with the impact of the virus. By being able to deal with this day and night using an AI assistant, customer requests were dealt with faster and with lower or zero wait times. Average Handling Time (AHT) following handoff from the AI bot to human agents declined significantly as customers had already had some queries answered or the bot had gathered enough details to allow faster resolution by the human agent.
   2. Deflecting Service Requests Away from Human-centric Channels: A new wave of deflection, beyond pure call deflection, is emerging to deflect customers from the traditional phone, email, and live chat channels and encourage them to engage with a virtual assistant. As a consequence of the Covid crisis, this deflection model has taken off and is proving to be highly attractive for businesses as it offers greater opportunities for automation. Many online businesses use email or contact forms as a primary means for customers to reach out to their service centers but this is not really an efficient channel to drive service traffic to when you could instead offer a bot assistant that pops up with a prompt or avatar to “Chat with Me”, letting the AI bot take over and deflect the clunky inbound emails and form-fills that have to be sorted and answered by humans. The same applies to Live Chat. Deflecting customers who may hit live chat to firstly engage with an AI assistant provides the best of both worlds – the AI can do whatever it is trained to do but can always be designed to hand off to Live Chat if needed or requested by the customer.
   3. Automation of Common Service Interactions: An AI assistant can replace human agents on routine customer service requests, automating underlying tasks either completely or partially. Instead of incurring the high labor costs associated with a human agent dealing with a customer on routine requests, why not have a virtual agent take care of these but still offer a path to hand over to a human if the issue cannot be resolved by the AI.  Deploying an AI virtual assistant service doesn’t mean that human intervention is eliminated. The virtual agent just takes on the tasks that they can easily handle and automate while the human agents deal with more complex customer issues. This helps drive up self-service levels for a business that not only is attractive for customers but also offers more cost-savings for the business. Self-service is a great way to reduce the heavy load on a customer service center and ensure that users aren’t left hanging or are dropped. The use of AI in customer service also allows human agents to become more productive as they are less inclined to have to deal with routine or simple requests. All business metrics benefit; the cost of service delivery reduces, agent productivity and satisfaction levels increase, abandonment rates and dropped calls decline, and customer satisfaction CSAT and NPS increase.
   4. Collaboration between Virtual and Human Agents: Contrary to some opinions, deploying virtual assistants doesn’t make human workers redundant. This has been the thread through much of this blog and in the previous three scenarios. In an ideal situation, the virtual and live agents work collaboratively so that the AI handles many of the routine service interactions,  either fully automating them or partially automating before handing off to a live agent. As such, the AI assistant becomes integral to service operations and to the systems that are used to access customer and business information rather than acting as standalone workers. During a live chat session, for example, an AI virtual assistant can be used to assist the human agent in resolving the customer issue faster by understanding the customer queries and providing quick access to information and making suggestions for the next actions that can be popped into the agent’s screen or in the customer’s message window. This is really a highly collaborative bot-assists-agent model. Alternatively, the bot is integrated with the live chat system so that it can automatically hand over when it can’t meet the customer’s needs i.e. a human handover model where the AI understands its skill limitations and duly passes the baton to the human.