CloudApp receives investment from Adobe & launches plugin for Adobe XD CC!
Read more

Using AI to Automate Customer Service

Sarah Blackstock

Automated customer service—through chatbots, self-service support, and other uses of artificial intelligence—have become a real factor in assisting customers in the past few years and show no signs of going away. AI can be used alongside agents to handle the more straightforward issues, help your customers find the right support doc, and to triage and sort the trickier questions, so your team can spend their time answering support tickets instead of organizing them.

Using automation in your support flow, you can:

  • Assist more customers than you can currently help with human power alone.
  • Shorten your team’s time to reply and resolution.
  • Reach out to customers proactively when they encounter a problem in your system.

No matter if a ticket is resolved using AI or an agent, it’s essential to craft accurate and personalized responses. According to Conduent's State of Customer Experience 2017 Tech report, 68% of customers expect tailored communications, and 18% of those customers expect the communications to have an advanced level of personalization. Sending replies that are missing the human element is no longer enough.

By answering more quickly, helping in more instances, and being proactive, you’ll provide the kind of support people tell their friends about and encourage a more loyal customer base. If you can master all of that, plus keep your support feeling human and accessible, you’ll be serving superior customer service and are sure to stand out.

Using AI to Automate Customer Support

AI can’t yet match a real person in their ability to understand information, make judgments, express empathy, and assist in complex situations, so it’s important not to try to replace your support with automations, but rather to supplement it. By using AI to assist your agents in handling your support load, you free them up to take on the more difficult problems and provide the human touch when users need it most.

Many current support desk tools include AI or can be integrated with it, including Zendesk which offers AI ticketing and a chatbot. Intercom also offers a Chat Bot option. Replying to tickets using automated replies can be especially great for things like billing issues and account changes your customer can solve on their own if they know where to look. On the other hand, you’re likely to need humans to tackle the more technical issues or the bugs a customer might stumble upon and have no way to resolve themselves.

Finding Issues to Automate

When looking for ways to use AI to assist customers on your behalf, there are a few things to consider:

  • Which questions are repeatedly asked and receive the same basic answer from your agents each time? To spot them, try creating a set of tags agents can use to tag tickets so you can take a look at each tag and pull out the most common issues.
  • Is there an issue people tend to write in to ask for help with but can handle themselves if they knew where to go? A simple automated answer can help get them there.
  • Which questions are asked in a ticket but are answered in detail in your support documentation? Write an automation to point them to the correct doc.
  • Are there questions that require a human, but can be tagged and sent to the appropriate queue by an automated system? Save you agents time, and use automation to do the sorting.
  • Check your support doc stats. Are there pages in your documentation that get a lot of views? Since the stats indicate a lot of people are wondering the same thing, having an answer ready for your ticket queue is a quick win.

Once you’ve identified what to automate, you can craft replies bots will use to point customers to the appropriate support doc. Some automation options can also tag and sort tickets, even if it can’t reply to them, allowing your agents to stop wasting time on that and focus on providing quality customer support through their customized answer.

Personalizing Customer Support

In the rush to automate and get more done with fewer agent hours, it can be easy to lose the human touch. However, tech—even the automated kind— has become more personalized than ever across the board, so it’s important not to let that happen. Allison Pickens, Chief Customer Office at Gainsight explains:

“Consumers expect hyper-personalized communication, and timely and relevant engagements. Google tells us how long our drive to work will be. We shop our favorite stores online, and we’re delighted to see items we previously browsed. We drive by a new restaurant, and we’re invited to come inside with an appealing offer.”

If one of your customers is used to that level of personalized detail, reaching out to your support and finding an overly generalized, non-human interaction will be a disappointment.

Combining Personalized Customer Support and Automation

Even when letting a bot send answers for you, it’s important to sound human. Whether you’re crafting predefined responses for your agents to guide their responses or for AI to send, be sure to:

  • Personalize your greeting. Use a customer’s name when you can. Signaling to a customer that you recognize them as a human forges a better connection and can help a frustrated person remember your agents are human too.
  • Avoid being stuffy. Don’t let the need for these responses to be widely used mean you use language that’s overly formal. Aim for professional, but approachable and human. Read the responses aloud to make sure they sound like something a person would actually say.
  • Keep it simple. Just because you won’t have to type out a big explanation over and over anymore, doesn’t mean the message should be extra long. Provide the information needed and that’s it. If possible, include screenshots or short videos to explain tricky issues better. Beyond that, include a link to a support doc offering up more details just in case.
  • Aim to handle issues in one reply. Whenever possible, answer a question the first time you reply. Every additional message has the potential to increase frustration. Remember the person on the other side of the support ticket and make sure you’re not wasting their time.

Have Your Agents Craft Personalized Responses Too

If you’re using AI to automate replies, you’ll still need your real human agents to cover the queues too. AI isn’t able to match a person’s empathy or understanding of more complex issues, so while it can help clear out some of the simpler or repetitive tasks, the people powering your support team will make their own impact handling the higher-level tasks.

Whether you’ve got tiered support that results in shuffling customers between agents, or expect users to go from bot to human during the support flow, be sure they’re not stuck saying the same thing over and over before getting the help they need. Seventy-two percent of customers blame their bad customer service experience on having to explain their problem to multiple people, according to a survey by Capterra. If you can simply eliminate that pain point, you’ll be doing better than many other operations.

To do that, consider customer experience pro Blake Morgan’s advice:

"Trust is the cornerstone to all customer experiences. It can’t be built in a day, but it can be destroyed quickly. Focusing on each customer individually instead of rushing to get things done can help create a long-term relationship of trust."

If a customer reaches an agent in need of trickier support, encourage your team to go the extra mile. They’re not a bot, so they shouldn’t sound like one. Move past pre-written responses one simply sends—automations can do that—and utilize your people power for the truly personalized, detailed work. This is not only more helpful for your customers, but also more fulfilling work for an agent to perform day after day.

To take your support to that next level, the key is personalization. If a customer is stuck and unable to figure out how to use a certain area in your product, why not record a short video or GIF and add it directly into your help desk message to show them the way?

Or when you’re chatting with someone who seems like they’d benefit from a visual reference to help them over and over, why not take a screenshot and annotate it so they can easily take a look and get back on track anytime? Do what you can to prevent a customer feeling like just another number. This kind of support stands out and will not only be appreciated, but also earn the type of customer loyalty that’s impossible to come by any other way.

Conclusion

No matter how you choose to handle your support queues, personalization is key. By creating customized messages, you can make automated replies for basic issues feel more human. For the trickier ones, you agents can use personalized language to defuse frustration and make an impression that improves customer loyalty and their happiness with your product or service, even when they hit a snag.

About the Author: Sarah Blackstock

Sarah Blackstock is a freelance writer specializing in support and technology, and former support pro at Automattic. When she’s not renovating her house in Dallas, you’ll find her baking in her (new) kitchen or reading romance novels. Find her on Instagram: @sarahblackstock.

Read more from the CloudApp blog — Inside The Workflow

Join the Discussion