GetJenny Blog

What do Customer Service Chatbots actually do?

Written by Colm Ó Searcóid | October 16, 2019

Customer Service Chatbot use is growing exponentially. By 2020, 85% of customer interactions will be handled by software. Why?

People tend to ask the same thing again, and again. And again. 

For customer service agents this is a very familiar story. Learning the copy-paste shortcut is first on the agenda in many customer service teams and it's the path to an almost robotic data-entry position. Service agents hired for their people skills and ability to solve issues are being overrun with simple queries a bot should do. 

 

Customer Service Agents Need Time to Do their Job

Customer service is the primary touchpoint for customers to interact with your company on a conversational level. For every step in your customer journey, every personalisation of mass communication, this is the most personal to a customer. In part, it's usually because they are having an issue with your brand.

On the flip side, customer service agents are overworked. In an average support team, each agent could be handling separate conversations with 4-5 customers at the same time. 

Have you ever tried having different conversations with 5 different people at the same time? Did it feel like you connected on a meaningful level? Of course not. 

This is where many customer journeys break down. Quality of support is stagnant when it has the potential to grow. The undeniable solution is chatbots. 

Customer Service Chatbots take the questions people ask 70-80% of the time and resolve them so that human professionals can focus on fulfilling the more complex needs of the remaining 20-30%. 

 

What does this mean for customer service agents?

Agents focus on complex issues not normally handled by chatbots, where a customer has a serious pain point that needs resolving. 

GetJenny case study with the If Insurance Company

This might be a bug in your website, an issue with payments or delivery, or problems with the product. These high priority issues can have a significant effect on customer experience and are the types of issues people write reviews about. 

As customer service chatbots handle the common questions, agents can focus on these critical tasks:

Get Bugs Fixed

While chatbots resolve common issues, agents can address the annoying cases where customers encounter bugs in your service. They can spend the extra time troubleshooting and verifying these bugs so that product development teams have a better idea of the issue and are able to fix it more effectively. 

Provide a more personal experience

Since agents aren't overloaded with requests, they can focus on the customer — their needs and expectations. They can give every customer with a complex issue the feeling that your brand is on their side, resolving their concerns. 

Conflict Resolution

When customers can’t use your product, it becomes a major pain point in their experience with your product. These are incidents where they need to feel like the brand is on their side, and fixing the issue. 

When these situations come up, customers are at that point where they might consider leaving your brand. Poor customer service costs businesses around the world $338.5B every year. And it costs 5x as much to attract a new customer than to keep an existing one.

It's a no brainer that your team should have the time to keep your churn low and your customer lifetime value up. Customer service chatbots give them that time. 

Improve Product Development

Every great product is developed through feedback. Dissatisfaction is as important as satisfaction when it comes to improving your product. As evidenced by Volvo and HELIX, the feedback loop provides a process for continuous product development. 

Customer service is one of those channels. Use the time saved by customer service chatbots to gather opinions and improve your cross-communication in product development. This improves your product and makes it customer-centric. 

Chatbots go further than Copy+Paste

If all chatbots did was replace copy and paste, that would be kind've boring, right? Augmenting CX with chatbots is not just about replacing robotic work, it's about making robotic work smart:

Personalisation

Integrate your CRM with your chatbots. Bring them personalised experiences that go beyond knowing their names. Know the products they're looked at or bought already, and hone your support for those particular products. For example, in insurance:

"John visits an insurance company's website, asking about a new policy. A customer service chatbot could look at John's information about his existing policies, and explain that he's eligible to discounts for his new policy because he's already a customer."

Author's Note: Personalisation creates an extra layer of intimacy between customers and brands. Developing services that personalise a customer’s experience requires handling of sensitive data, so ensure that your services are designed to protect your customer’s privacy, such as providing an enhanced experience only when they’re logged into your services. 

Checkout Recommendations

Upselling at the checkout becomes easy. Chatbots can recommend popular products based on the cart items, or for example, companion products, like batteries for that toy or phone cases for the new phone you're buying. 

Feedback Gathering

Customer service chatbots help you accurately measure your NPS or CSAT numbers. Any interaction with an agent or bot can be followed up automatically with a survey. It becomes a mainstay at the end of every chat touchpoint. And with 85% of customers interacting with software for their support by 2020, you must consider effective, automated feedback gathering options. 

How to use Customer Service Chatbots

Customer service chatbots work in two primary ways. With and without Live Chat. 

 

Without Live Chat

If your team doesn't have a live chat customer service channel, a simple widget is a great way to start using chatbots on your brand's website. 

A chatbot widget demo from GetJenny

A simple widget chatbot is taught the common questions and answers normally asked. It examines each inquiry and gives an answer. If it’s unable to answer the question, it explains this to your visitor and, if taught to do so, gives them an alternative method for finding out what they want to know. 

These chatbots work alone, so there is no option for transferring the customer to a human representative automatically. If you don't have Live Chat, this is not really an issue, but it does give you a valuable tool when building an omni-channel customer support team.

Escalations: Something the chatbot can't understand

Whenever a chatbot runs into an issue it can't understand, it flags the question as an "escalation". This makes a new record in the chatbot's database. 

Human agents check the escalations and teach the bot to answer the question correctly the next time. 

What's so good about this anyway?

Generally, repeated questions amount up to about 80-90% of your chats. Each resolved escalation brings you closer to automating this figure with chatbots. This gives your team more time to handle the critical remaining 10-20%. 

Chatbot-First Strategy

Use a standalone chatbot as the first step to improve customer experience. If you're thinking of investing in Live Chat, chatbot escalations give you a valuable insight in estimating the size of future Live Chat teams, the times they should be live and the types of complex questions they would handle. 

 

With Live Chat

Increasing the efficiency of your team is as easy as introducing chatbots into your support channel mix. 

A customer service chatbot working with a Live Chat team uses escalations to an extra advantage, and provides a seamless support transition for your customers. 

How Escalations work with Live Chat

When chatbots work with Live Chat, and there's a question they can't answer, they still go into escalation. But now the option exists to transfer to a human representative if there's one available. 

Demo of escalating to an agent from GetJenny

The chatbot transfers all the information to an agent, and this gives the human experts a foundation to resolve the issue more quickly. 

Why this helps.

Even when a chatbot cannot address the issues alone, it can still gather useful information. A human agent is then ready to resolve the issue without needing a long list of qualifying questions, since most of this can already be done by the chatbot. 

 

Bot Whispering

Customer service chatbots learn by examining keywords or questions received and associating combinations of those keywords, or the questions themselves, with answers they can give to your customers.

The process of teaching chatbots these keywords and questions is known as bot whispering. A bot whisperer is a new position for a selection of your agents. 

Bot whispering shouldn't be rocket science. Teaching a bot is not about programming or coding, it should be easy and intuitive. You should look out for a chatbot solution that:

  1. Has an easy-to-use interface. Bot whisperers administer the bot easily and effectively through the chatbot’s configuration interface. The need for training should be minimal.
  2. Is easily accessible. It should be as simple as logging in and making changes to the bot. 
  3. Makes desired changes instantly. Bot whisperers should be able to make changes to the chatbot in real-time. If the chatbot makes a mistake, it should be possible to fix it immediately. No coders involved. 
  4. Provide chatbot analytics. You should be able to see how many questions your chatbot resolves so that you can measure its effectiveness on your overall customer service offering. 

 

Bot Whispering and AI

Bot whispering with keywords and questions is a great start to handling up to 80% of your customer service issues via chat widgets. But it's not only about using the keywords set by bot whisperers. 

Natural Language Processing, synonym and common spelling error detection is a key to quickly scale your chat service offering. Chatbots that detect these will quickly expand their knowledge based on the customers input and the answers taught to them by bot whisperers. 

This makes it easy for chatbots to quickly scale their answering capabilities. It also makes it possible to learn to associate new questions customers ask with existing answers. 

Please Note

Although chatbots can learn quickly through this method, they need to first; have a base of knowledge to work off, and second; it's important that customer service chatbots have certain limits. 

Human agents should always be the line of defence. AI, although powerful, is still a relatively new technology. Mistakes can, and have been made. Protecting your brand integrity means making sure that humans always have control over your chatbot. 

 

Chatbot Do's and Don'ts

  1. Regular Maintenance - keep your chatbot oiled by assigning a bot whisperer to go through escalations and add or edit unanswered issues. This way, your chatbot continuously learns to automate more issues.
  2. Be Transparent - people expect brands to be honest with them, and they'll see through a bot fairly quickly. Stay honest about the fact that your customer is talking to a bot. Like the case of Teppo, the Turva chatbot, they understand that machines make mistakes and don't get irritated if their issue needs to be transferred. 
  3. Privacy Matters - ensure that your chatbot doesn’t handle sensitive information, if it isn’t meant to. Be clear to all your visitors about how information is used by the chatbot and give examples of specific information they should not provide. 

 

Conclusion (TL; DR)

Customer service chatbots are a powerful addition to your brand's customer-facing image. They add an automated interactivity that's sorely needed in the fast-paced, competitive landscape where experience is king.

Whether you implement a chatbot into your customer service offering with or without live chat support it's essential to remember:

  1. Use a service that's easy to install and easy to operate. GetJenny's chatbot solution was found to be 10x easier to use than its competition in a public tender comparison. [partnerships in practise]
  2. Be transparent about using chatbots. Don't think that machines are smarter than your customers, because they're not. 
  3. Maintain your customer service chatbots regularly. Assign a bot whisperer to go through escalations and teach your chatbot to serve your customers more extensively. This is how you hit an 80% automation rate. 

GetJenny offers a customer service chatbot and training solution catered to your brand's needs. Whether you have a live chat team or not, it's easy to bring smart customer service automation to your brand. Talk to one of our business development managers today.