4 Myths About Chat Support. AI to the Rescue?

This article was originally published on CustomerThink on February 13, 2018. Click here to read the original.

A big reason chat has taken hold in the market (setting aside customer preference) is that it promises reduced cost per contact because agents can handle multiple chats concurrently. For the purpose of this article, when I say “chat” I’m referring to all live messaging channels including Facebook Messenger and SMS.

Speaking of cost per contact, I didn’t even know what it was until I read Call Centers for Dummies as a new support leader. I quickly learned that cost per contact can be difficult to calculate (typically total cost of support divided by total contacts handled) and even trickier to break down by individual support channel. That doesn’t stop companies from touting chat as a less expensive channel than phone and email.

Rather than venturing to break down my own cost per contact for each support channel, I’ve used a guideline from Contact Centre Helper that estimates chat support is 17-30% less expensive than other channels.

Myth #1: Agents can handle other channels while doing chats

When I first added chat to our suite of support channels more than a decade ago, volume was low and I asked my agents to handle chats “in between” phone calls, which inevitably meant “during” as well. After all, we were doing emails in between calls with good success. That practice promptly ended when I tried taking chats and calls at the same time. The result was two mixed up conversations, and a terrible experience for both customers, not to mention a serious security risk!

From the standpoint of support leaders and workforce managers, chat can be tricky to plan for. It’s only normal to want to add it to the plate of your existing team without making significant workforce changes. The reality is that you’re going to have to dedicate staff to the channel accordingly and plan on them not answering phone calls.

Recommendation: Don’t ask agents to handle more than one live channel at a time. This means you shouldn’t blend phone and chat but you can blend chat and email or phone and email assuming there are some gaps between chats and calls.

Myth #2: Chats are more efficient than phone calls

I can remember getting engaged with a support team a while back and quickly learning that they had decided to only handle one chat at a time. When I was running a support team, we found that while our average handle time for phone calls was typically 8-10 minutes, our handle time for chats was more like 10-15 minutes. That means a single chat takes longer and therefore is more expensive than a phone call. If you plan on taking one chat at a time, you’re better off dropping the channel altogether if the handle time is significantly longer than phone.

The real efficiency of chat is found in handling multiple chats concurrently and that assumes there are significant gaps in conversations with customers. Granted, there will be times where agents are in a really involved chat or perhaps they need to call the customer for easier troubleshooting. In the spirit of great customer service, it’s important to give your agents the latitude to move the conversation to a better (more efficient) channel at their discretion.

Recommendation: Plan on your chat agents handling more than one chat at a time to maximize savings.

Myth #3: Agents can handle many chats at once

While this one is technically true, there’s a balance that needs to be struck. As I’ve already said, too few chats and it’s probably not a viable channel. Too many chats, however, and things start to come apart at the seams. I surveyed our programs at FCR and found that most teams are handling between two and three chats at a time with very few going beyond that. Many of our teams at FCR do start brand new agents on one chat and then ramp them up to more over the course of a week or two as they become proficient.

It’s important to understand the complexity of the chats for your own company before determining how many your agents can handle. For example, the more windows agents are required to have open for these conversations, the more likely there are to be mix ups in conversations. Also, the more conversations agents have going on at once, the more likelihood of long delays between messages for certain customers.

Recommendation: Agents should be handling multiple chats concurrently but it’s important to strike a balance. As you’re reading through chat conversations, if you see customers saying “Hello! Is anyone there?” or something similar, your agents might be handling too many conversations at once.

Myth #4: Any great customer service agent can do chat

When I began staffing for chat, I initially rotated quite a few of my team members onto the channel because they loved the change of pace from phone calls. What I learned was that there was a much wider variance in average handle time on chat in comparison to phone. I had some agents averaging eight minutes and others averaging twenty.

I decided to have all of my agents take a typing test and sure enough, my more efficient agents typed ninety or more words per minute and our least efficient were below sixty. From that point, we changed our hiring profile, had candidates take a typing test, and filtered out anyone who couldn’t type more than sixty words per minute. Out of our existing team members, the faster typists spent more time doing chat and the others handled phone support instead.

Recommendation: Not everyone can easily do chat. Hire great customer service professionals who can type fast to maximize efficiency.

How AI Can Help

Chatbots and macros are among the more popular ways to make chat more efficient but I have a couple thoughts here. First, chatbots still seem to be a good thought but for any company that’s not a large enterprise, it might be a challenge to find issues with enough volume to justify building a bot. Many tools exist to present self-help options to customers before they submit a chat and that’s where I’d start.

Second, macros (aka canned responses) are great for efficiency but customers can still smell “canned” a mile away, especially when they aren’t tailored to the customer’s specific situation before sending. One possibility I’ve seen recently with companies like Digital Genius and TrueAI offers better macro suggestion for chat. Rather than giving THE right answer to agents, they suggest a handful of responses that have worked for that scenario in the past and resulted in high customer satisfaction.

With this model, humans work alongside the AI with humans only answering the questions that the AI is unable to. This promises significant reduction in the time to train new agents and increased consistency and accuracy of responses while not losing the human element.

Conclusion

On a recent call with a vendor, my cost per contact brain asked if AI macro suggestion would enable agents to handle more chats concurrently. He was quick to correct me, stating that it actually enables agents to handle chats at a faster pace without ratcheting up how many they need to work on at once. Any increase in productivity without ratcheting up the stress is a win-win.

Whether you’re ready to adopt an AI solution or not, chat support offers significant savings for your support operation right now if you do it right. If you’re hiring customers service pros who can work at the right pace, dedicating people to the channel, and finding a balance for how many conversations agents handle at once (more than one), you’re well on your way to reaping those benefits.