The Playbook for Customer Service/Support AI

WRITTEN BY PUNEET MEHTA

The Playbook for Customer Service/Support AI

On the surface, applying chatbots as virtual assistants for agents and directly for customers, and artificial intelligence (AI) in general, can seem like a daunting and overly complex undertaking.

There is serious and creative design, implementation (including problem solving) involved, along with – and most critically – the training of not only the internal teams but also the agents and yes customers to make these solutions happen.

After all, when it comes to delivering customer service, aren’t human agents sufficient enough? Isn’t the human element the most important part of customer service?

Well, yes, but times are changing. We simply have to look at expectations and preferences of customers today: which have evolved tremendously over the past several years.

Many have grown to welcome new technologies – a 2022 report from Zendesk found that 66% of customers applaud AI for making their lives easier – while research from Salesforce found that 60% of customers are open to the use of AI in customer engagement, including 66% of millennials.

Many agents today are feeling the pressure to resolve tickets at speed…

We also have to look at the agents themselves. How can their lives be made easier, amid an era of rising customer expectations and more and more tickets to close? It’s also a rocky labor pool – at around 45%, support agents have one of the highest turnover rates of any industry.

Many agents today are feeling the pressure to resolve tickets at speed and deal with impatient and irritated customers – a new report from Netomi found that 73% of customers have experienced instances in which agents have been rude to them, and 38% have experienced an agent becoming upset.

Needed: A Strategic Approach

In this environment, many companies are adopting AI-powered chatbots to boost business efficiencies and enhance the overall customer experience.

However, simply introducing a new technology for technology’s sake is not the answer. Instead, businesses need to follow more of a strategic approach to ensure that the agent and customer experience is uplifted.

Today, companies have no choice but to adopt AI, but they have to do it right. The band-aid solution for many was implementing first-generation bots, yet those often did more harm than good, resulting in higher costs and increased customer dissatisfaction by driving them to other channels to get their issues resolved.

Businesses need to follow more of a strategic approach…

A well-designed and carefully planned chatbot strategy is key – for instance, ensuring the dialog is adapted to the channel (i.e., short messages on chat) and offering a seamless escalation path.

When it comes to chatbot adoption, here are some tips on what to do right, as well as recommendations on how to maximize the benefits of chatbots.

A. Train the AI-powered bots to work alongside human agents, acting as sidekicks.

Virtual assistants, also known as virtual agents, need not replace human agents, rather the two can work hand-in-hand as the ultimate high-performing team.

These AI-powered applications help to automate many of the high-volume, repeatable tickets and queries that support agents receive on a frequent basis (such as airline baggage policies and subscription renewals).

By eliminating mundane work, agents’ jobs are more fulfilling as they focus on high-value issues that require skills related to creativity, problem-solving, and critical thinking.

In addition to resolving repetitive queries with customers directly, AI virtual agents can also help human agents with more complex tickets by doing the “prep” work, such as gathering flight booking information or order details from the customer, before intelligently routing customer queries to the right agents.

The bots can also capture information and contextual data from back-end systems such as order management or CRM platforms on behalf of agents, as well as suggest responses, which the agents can then accept and send or edit. Think of it like an administrative assistant who takes notes at a board meeting to share with other employees.

Preserving the human element in customer service is key, which is why it is essential to always offer an escalation path to human agents, based on topic, user sentiment, customer profile, or when a customer asks to interact with one.

If a customer is becoming more and more irritated, for instance, a human agent may need to intervene. Similarly, if an issue is highly complex or involves sensitive information, a human agent would need to take the lead to ensure the best end-user experience.

B. Create an engaging and helpful chat experience, while providing customer self-service options.

Lengthy interactions with a human or virtual agent are not always necessary – sometimes customers just need to be pointed in the right direction.

Customers today like searching for answers and coming to conclusions on their own. Just think of everyday activities such as online banking and self-checkouts at grocery stores.

Serving as tour guides in the customer journey, chatbots can seamlessly guide customers towards self-service resolution while making it conversational and engaging.

While many first-generation bots simply point users to help desk articles, AI-powered virtual assistants can create two-way dialogs to ask customers for exactly the information that is relevant to them. So, if a hotel guest is asking about a free shuttle to the airport, the assistant can confirm their arrival time to provide the most accurate information to that guest.

While knowledge bases are excellent backup options if there are topics on which an AI has not been trained, the chat experience needs to be engaging.

So, when designing dialog for chatbots, focus on the customer, and weave in statements of acknowledgment to build trust and make the conversation flow naturally (such as: “Got it. So, you’d like to return the sweater because it’s the wrong color?”)

C. Select the right channels, ensuring a consistent experience across them.

Customers expect support across channels – whether that may be SMS, chat, Facebook Messenger, email, or voice platforms such as Alexa.

Consider Starbucks with its mobile app, loyalty program , and order-ahead functionality, which enables customers to easily place their orders, pay, head to a Starbucks location and have their orders waiting for them.

Chatbots can aid in a similarly connected and consistent experience, allowing companies to engage customers across channels.

In doing so, it is also important to ensure fluidity across channels and make sure that the users can maintain their conversation history across all channels, and with a single customer profile. If a customer reached out via email in one instance, and Facebook in another instance, they should not need to restart the conversation.

Chatbots can aid in a similarly connected and consistent experience…

That said, virtual assistants need time to learn, and launching a bot on a channel that has not seen a lot of traffic will not give it adequate time to learn from real-life conversations. Therefore, it is important to start out on those channels on which your business already has a presence, and eventually scale to additional channels.

These are among the benefits of chatbots, and instances in which these powerful tools can serve as the ultimate sidekicks for support teams across industries – fintech, retail, travel and hospitality, and more.

Yet there is a time and a space for everything, and several instances in which chatbots might not be the right solution is when there are:

1. Reliance on keywords.

Many chatbot providers rely on keywords in order to trigger the right response. This is very limiting, however, and often leads to a poor user experience as the bot will get confused if it hasn’t been trained on the exact phrasing that the user says.

For instance, for “order status” a person might ask “when will my package arrive?” or “is it coming today?” If the AI hasn’t been trained explicitly to understand “order status” in these ways, it will not be able to answer correctly.

Instead, look for AI platforms that leverage semantics. These systems look at the context of the user’s interaction, their history and how words relate to each other instead of their dictionary meaning. This results in a much more accurate AI, less lift from the internal team, and a much better end-user experience.

2. Lack of ongoing training.

Even systems that leverage semantics are not “set it and forget it” type of machines – they need to continually learn, and they require training and human supervision.

Many companies have historical data which can be used to jumpstart AI training, but once it’s launched, it’s important that teams play an active role to optimize its performance over time. This includes tracking which queries are not understood by the bot and mapping them to the right topics.

For instance, if a customer asks a question about upgrading a flight in an oddball way and the AI doesn’t map it to the correct topic, the team can let the AI know which topic the user’s query should correlate to so it can understand it in the future. Supervision also allows support teams to uncover new topics that the AI could automatically resolve with zero-human effort to uncover even more ROI.

3. Lack of metrics and overarching strategy in place.

Consider what the overall goals of your business are, and how a virtual assistant can help you achieve those goals.

Are you looking to scale to new channels, drive more revenue or tackle tedious and time-consuming administrative tasks?

With any initiative, measuring the impact it has on business goals is critical. You’ll want to understand how the AI is performing, how many times it needs to escalate to a human agent and how it is improving over time.

Given that most conversational AI programs are launched to improve the customer and agent experience, it’s important to also track the results within the context of the larger organization. Is better support positively impacting churn rates? Is proactive care driving more revenue?

In conclusion, human-only customer service and support workforces are no longer the sole answer. Support teams have evolved to keep up – to deliver an omnichannel, personalized, proactive, and always-on customer experience, one in which AI-powered chatbots can play a leading role.

Puneet Mehta

Puneet Mehta is CEO, Netomi. He spent much of his career as a tech entrepreneur as well as on Wall Street building trading AI. He has been recognized as a member of Advertising Age’s Creativity 50 list, and Business Insider’s Silicon Alley 100 and 35 Up-And-Coming Entrepreneurs You Need To Meet.