I Hate Bad Bots! 5 Ways to Create a Better Automated Experience

I called my bank the other day and was greeted by the obligatory phone tree. I answered five different levels of questions before finally getting frustrated and asking to be transferred to an agent. When the agent “Karen” picked up the phone, she asked me the same series of questions I had just answered with the IVR. Nice. After we got through that, she determined that I should talk to someone else and transferred me. After much ringing, an agent picked up.

“Hello! This is Karen. How may I help you?”

“Um…we just spoke. You were going to transfer me?”

“Sorry. Let me transfer you.”

A lot more ringing.

“Hello! This is Karen. How may I help you?”

“Yeah. It’s still me.”

“Sorry. Let me transfer you.”

More ringing. Then I’m transferred into the original phone tree. Aargh!

The balance between human and automated interactions is a difficult one for customer care centers. While using human agents is costly, over-reliance on automated systems can cost dearly in customer satisfaction. In fact, IVRs are the least popular service method, with 83% of consumers seeing no personal benefit to them.

Automation, when used well, should be about crafting excellent customer experiences. Period. Contrary to popular belief, automation does not exist merely to reduce staffing needs or to cut costs. Of course, good automation can indeed do those things over time. But if a business begins with the question, “How can automation serve our internal processes?” the battle is already lost.

With that in mind, here are some ways to tame those bad bots:

1. Find out what the customer wants from automation. Then figure out how to do it.

According to Forrester Research, we are now in the age of the customer and the businesses that survive are making customer-centric decisions. With that in mind, I am surprised how many business owners and customer service managers I talk to that have not begun figuring out customer needs.

This goes back to taking the customer perspective. When considering automation, ask yourself, “What is the primary issue I am trying to solve?” If your answer has something to do with user experience, you are on the right track. However, if your primary reason has to do with “reducing staff” or “standardizing data entry systems,” now would be a good time to switch perspectives and find out exactly what your customers want.

There are a number of ways to do this. You may use customer satisfaction questionnaires or verbal surveys. You can also perform diary studies or shadow a day in the life of the customers. Creating personas is also a time-tested strategy for getting to know your customer.

While human-human contact might seem like the gold standard, you may discover that many customers would prefer to have a human interaction for some tasks (like troubleshooting) and an automated interaction for other tasks (like confirming an appointment). When choosing where to automate, look for places where it would simplify a known task or where the end user finds human-human interactions unwieldy or frustrating.

2. Whether using text or phone for automation, always have a person overseeing the process.

Until artificial intelligence is more developed, an automated system cannot be truly smart or adaptive. What it CAN do is collect data and find patterns. What it CANNOT do is find meaning and solve problems creatively. For that, you need people.

Let’s say you go to a hospital emergency room with a tightness in your chest. The triage nurse evaluates the severity of the problem, enters your data into the system, and figures out where to put you. You are then sent to a room and hooked up to monitors which display your heart rate and blood pressure. A doctor comes in, asks how you are feeling, examines you, looks over your medical history, checks the monitors, and determines a course of action.

This is the reality of how humans have adapted and evolved to interact with machines. The machines in our hospital scenario, including the medical history software or the monitors, collect and display data, but the triage nurse and doctor help interpret and make sense of it for you personally.

In business, this means that humans are needed to triage and solve customers’ problems, while using automation just for data collection. Sadly, many contact centers fail at this—creating the “bad bots” that customers have come to loath. How many of us have been caught in the phone tree from hell or wasted time getting nowhere on an interactive troubleshooting webpage?

This brings me to my next point…

3. Use a combination of human and automated interactions.

Not only should people oversee the automation processes, but an awesome strategy is to let humans and robots tag team on the same interaction. Here is an example using ITR:

1. A customer texts the support center to say his router is dead, the light is blinking, and he is stumped.
2. The agent texts back: “What is the make and model?”
3. The customer sends back his answer.
4. The agent replies: “That’s a common problem. Here is how to fix that.”
5. The agent initiates an automated interactive script and moves on to helping other customers. If a customer reply goes off script, the ITR can then seamlessly pass the conversation back to the human agent.

Not only does the customer get a live agent during crucial moments in the interaction, but the agent can move between customers during the periods of scripted automation. In this way, the agent can be like a doctor who moves easily between patients, stopping by each one to assess the situation while letting the monitors do the rest.

Many contact centers rightfully want to provide a human touch, but the employees with the greatest expertise waste their time doing data entry and relaying the same scripted information. By teaming up machines and humans, however, a company can make the customer feel happy and cared for, while maximizing the time of these highly-trained agents.

4. Never ask for the same information twice. Instead, create a seamless experience between channels.

No discussion of automated channels would be complete without a mention of seamless integration. According to current social norms, if I call an associate, then follow up with an email, then send a text to ask her about an upcoming meeting, I assume she is aware of the information in all three channels. 

Unfortunately, in many current automated interactions, customers are left wondering where all of their information on different channels goes. A common example is the customer service agent who asks for information the customer has already input into IVR or ITR. One of the surest ways to get a conversation off on the wrong foot is to make a customer repeat themselves. This breaks the rules of what we have come to expect from the human-machine interaction.

Going back to our hospital scenario, how irritating would it be if the doctor came to your room to ask what your heart rate was? You would rightfully wonder why you are bothering to wear a heart rate monitor if the doctor isn’t willing to walk over and look at the display.

In the same way, when a customer enters information into an IVR or ITR system, they expect that the agent is looking at that information. Otherwise, why would the company have asked for it? A customer should be able to switch between channels without anyone losing track of where they are in the conversation.

5. Use data and customer history to make automation more intelligent.

Even though automation is unintelligent by nature, it can be made to appear brainy and personable by mining customer data and conversation history. For example, once a specific customer is identified through input data or caller ID, the IVR or ITR software can make an educated guess as to why that customer is calling and present it as a first option:

For IVR: “Hi! If you are calling to check the status of your router repair, press 1. For all other inquiries, press 2.”

For ITR: Hi [name]! Please select from the following:

1. Status of my router repair
2. My account
3. Troubleshooting
4. Other

As a customer, one of my least favorite parts of a service call is the part where I have to start at the beginning and explain to the agent who I am and why I am calling. I get such a huge feeling of relief when I realize that the company already knows who I am and is proactively solving my problem.

In this article, I have mainly just scratched the surface. As the relationship between humans and machines continues to evolve, there will be even more elegant ways that they can work together for customer care. The bottom line is that automation can be a great tool when combined with human interaction to delight customers. But can turn on you if used at the expense of customer satisfaction.

Nice robot.

More Resources

• Webinar: 5 Steps to Modernize Your Contact Center
• Research Report: The Multichannel Agent: A 2014 Contact Center Roadmap
• Educast: Proactive Engagement in the Multichannel World
• Training: ICMI OnDemand All-Access Training Pass