How To Elevate Customer Service With Effective AI Use With Marc Bernstein
Most customer service integrates AI into their processes to speed things up and address a bigger number of concerns every day. However, many people are not satisfied with just talking to chatbots only. Chad Burmeister sits down once again with Marc Bernstein to discuss how they found the right balance between AI and human intervention at Balto. Marc explains how AI should never overtake human responsibilities completely but only assist in filtering, determining, and analyzing client questions, resulting in better (but still humane) conversations. He also shares why they named the company after a Siberian Husky that became a hero during the 1920s, connecting the dog's brave act with AI-human relationships.
Listen to the podcast here:
How To Elevate Customer Service With Effective AI Use With Marc Bernstein
I've got a guest coming back with us, Marc Bernstein. I have talked to Marc multiple times because I am so intrigued and enthused by what it is they are doing. They are bringing AI to conversations to enable people to have better conversations. We all need the training wheels from time to time when riding a bike or talking on the telephone. Salespeople powered by AI and guides outperform people who do have those kinds of guides in a very large way. We are looking forward to drilling down, Marc. Welcome back to the show.
Thank you so much, Chad.
It is great to have you here. There is a friend of mine, Jason Adams, who is in Dallas. He said, "If you were to summarize in a sentence what this company does, I have heard you say yours before better than most." How would you summarize what Balto Software does in one sentence?
It analyzes what sales and customer service people say in real-time and gives them guidance on how they can be as effective on the phone as humanly possible.
I love the word humanly in there because that is what good AI does. It makes us more human, not less human.
It is so funny, Chad. I will often do a little play on words and I will say, "It can make sales and service people more effective than humanly possible." It is this idea of a synthesis between, "What are people good at? What are machines good at?" Both machines and people have their strengths. If you are using AI as a tool to achieve an end rather than something to replace your existing operation, you are enhancing the capabilities of people. People have already created some amazing things like we created the pyramids. What is the next thing that we can create in the sales and service world? The opportunities are endless.
AI must bring people together in moments of coaching that they truly love, with everyone at their peak to the end.
I had a professor at my house, Dr. Jim Wilder. He wrote multiple books, including The Solution of Choice. In this book, he is considered a neurotheologist. He said that, "In the Old Testament, there are 614 simultaneous laws that one must consider when they are making a decision and a choice. There are more choice options in the entire known universe times 2, times 3.14." That means choices are hard. The best way to make a choice is to think, "If the creator created the choice it made me then what would that person do and make at this point in time?" What is interesting about this conversation is that giving AI to people brings together all of the known information about a conversation. The customer has X number of users deployed and they have all kinds of information but they would never be able to go in and preview for a phone call. You are enabling, bringing that together to the forefront so they can have deep and meaningful conversations with the customers.
For anyone out there who is thinking about some real-time guidance that they are evaluating or considering, I will give everyone a purchase tip. One of the things that are most important is, when the real-time guidance is giving salespeople recommendations, it needs to give them options. It needs to say, "Here are five different questions you ask." Use your human brain to the best possible decision. People hate when AI recommends something that says, "This is the one way to do it. Do it this way right now," because they go, "It is not me. It is not who I am. I would like it a little differently."
When you give people options, they can synthesize how they want to approach the situation with what data says are the optimal ways to approach the situation and find that balance between them expressing themselves and doing things the way that they would without any guidance and data saying, "We have analyzed the numbers in the most effective way to have the customer conversation to ask this question." You got to get that balance.
It makes me think of all the years that I've managed reps. Now, I don't have the ability to be there and sit next to them in an office on a whiteboard. When I did have that ability, I would go up on the whiteboard. We would be on a speaker call with the customer. Often, the customer didn't know I was there as the manager. I would synthesize the information using my powerful AI and my brain and write a question on the board for the rep to ask. It was brilliant because you could mute the phone, talk to the rep and say, "You need to do it this way." I remember one time this rep was saying something that was not good. I paused it and went, "Stop. They just bought it. You need to understand they are ready to send you a contract." We unpaused the phone and we were like, "We will go ahead and send out the contract." He was about to step himself out of a deal.
Having the ability to have a sidecar of somebody there all the time answering the question, "What would Chad do? What would the best sales coach in the world do?" That is what Balto Software is all about and that is what I am enthused about what you are bringing. I have a personal soft spot because I was fired from my first sales job out of school. I look back and if I had guidance software like this, things could have turned out differently. This is not a risk to sellers. This is a help to sellers in a massive way.
I heard two things there. The first thing is your whiteboard story hit me. I think anybody who has been in sales or sales management can picture that whiteboard story and go, "Yes." I picture a time where I and a rep or me and the manager were together in a room and they were scribbling notes. I asked this question versus that question. We won the call or got close and then we celebrated. What is so important about that is that the whiteboard story is coaching the way that reps and managers want to be coached. It is not, "Listen to it after the fact and I will ask you questions through the Socratic method," while you are like, "Come on. Tell me the coaching advice."
You are like, "No, you got to figure it out on your own." That after-call coaching is a dynamic often, where neither party enjoys it. The rep is like, "Please, just tell me what I need to do better." The manager is like, "I have to spend all this time listening to the call, recording and grilling you." When you bring them together at a moment like that, that is the coaching that people truly love. Those are the moments where everyone high-fives at the end. That was the first thing I thought of, Chad.
I can remember that very clearly. That was one of those moments where I am like, "I know that person just learned and it is going to give them the why in the road. They will take that away from that interaction. Now, you can do it at scale across millions of phone conversations. Forget about it. I want to do that. I can't be on every call but now, I can be on every call." That is the beauty of what this technology does.
There was another person who called it. I remember Curious.io that was out. I had dinner with Sabrina in San Francisco. She said, "The types of tools that Marc and company are building are inside the conversation." There is inside and there is a post. She called it postmortem. "The call is already completed. If it died, you can't bring that call back to life. It is already dead." That is the beauty of bringing this AI to the forefront. Let us talk a little bit about how it works because I think people are getting the concept here, "But I can't physically be there on a whiteboard with somebody." How does AI perform this task, listen to the dialogue and give me the conversation of what I should say?
You can break it down into a very simple timeline or process in four parts. First, you got to grab the audio stream. You got to hook into the phone system and make sure that the audio is flowing through and that you are able to capture and analyze it. The second is you got to do real-time transcription. You got to turn the audio to text. It is difficult to extract sentiment from an audio stream. You have got to convert it into some other format. Some people try to analyze the tone but I always compare that to Charlie's parents and Charlie Brown. You can't figure out what was said. You can figure out the tone. You have got to convert it to something else.
The second step is going to convert it into text. You have this stream of text. and logic, "Was this You can look at the stream of text and say, "What is happening here?" Parcel out different words, phrases, sentences, context said before that? Was that said before that?" Assign labels or events and go, "A-ha, this was a pricing question, this was an objection, this was a qualifying question." You have your audio, speech-to-text and labels. When you find that moment, you go, "A-ha, it was a pricing question." The last thing is you do deliver the right content for that moment. The content is the questions that everyone wishes that people would say live on the call.
When someone asks about pricing, everyone has this set of best practices in their mind. They say, "Please, first ask, 'Did they like the product they saw? Is there anything else they would like to see before we move to pricing? Is there anyone else that needs to be involved in the pricing conversation?'" You want to have those events. They are paired with content. When you pair it with content, you are able to give a recommendation to the representative. Right there on the call, it will pop up for the rep and say, "Here are your pricing questions. Here are your qualification questions." That chain is a chain that is very difficult to complete from start to finish and Balto has been uniquely good at delivering on it.
The relationship between AI and humans must be about doing the things flawlessly that the other party cannot.
Share an example of maybe not the exact customer because they probably look at this as one of their most strategic weapons they could own so they don't want to share it with other people. I am assuming you can say, "CPG company with 100 people in a call center or something like that." What are you allowed to share? Let us talk through one of those case studies.
One of the use cases that are good for Balto is insurance. The reason is it is such a high customer lifetime value. If you get that interaction right, treat the customer, match whatever their needs are with a policy or text them and also is a reasonable cost then you have a happy customer for a very long time. We do well in this space. There is one insurance carrier for property and casualty insurance. They have about 2,400 people in Balto. One of the things that we do with new customers is we give everybody nowadays if you are over a $20,000 deal or something. For everyone over $20,000, we will do a controlled A/B split test, where half of your people will be on Balto and the other half will be a control group, not on Balto.
We give them almost a watered-down version of Balto. We call it Balto Light that does all the analysis of the call but gives the rep any help. There is no guidance. What we are able to do is compare the results between the Balto group and the non-Balto group. For that insurance care, what we did is we pulled their managers together and said, "What are the absolute best discovery questions that you notice gets right to the heart of the customer need?" They did this big brainstorming session and they came up with all the great questions. We said, "We are going to put all of those right in your Balto checklist.”
Now, front and center, every rep has all these questions right in front of their eyeballs so if they ever feel lost and they feel like they need to make a change. They can even compete on who is asking the most discovery questions because everything that the rep sees is also tracked in the analytics in the background. You can have people competing over who is doing the deepest discovery while they launch this process. After a little over 12 or 14 weeks, the Balto group at the end was outperforming the non-Balto group by 50%. Their conversion rates were 27% for the Balto group versus 18% for the non-Balto group.
I want to make sure that our readers understand what this means. I met with a marketing consultant. He asked me a bunch of questions about my business and about ten different items on his list. The one area that we lacked in is now that I have a team of five sellers. When it was me, my conversion rates were 30% to 50% of every sales call turned into a deal. Now, it is 10% to 12%. I view myself as a very good seller and sales leader. I can put it into slides and Salesforce with, "Here are the questions." I can monitor it through a post-call recording and put all the notes in Salesforce but if I don't provide real-time coaching, I will stay at the 12% conversion rate. I will never get the team up to 30% to 50%.
When I hear technologies like this, I am so excited because it allows me to scale the best practices I have learned for many years. Not just me and what is in my head but other people like Keenan of Gap Selling. Let us put the Keenan Gap Selling process into play when you are doing conversations. This is very important. Balto means to share with us. I remember watching a movie about Balto. I remember this dog running across the ice that was collapsing under him. He saved a lot of people, if I remember right. I think your Balto version of the software is also saving a lot of people. What is the naming of Balto after?
Thanks for asking. It is funny. One of our first tag lines was Balto Saves Calls. The Balto origin story is that it comes from a real dog in the year 1925. Disney made a movie about this. In 1925, there is this terrible outbreak of this thing called diphtheria, which we now get vaccinated for. In the Northwest corner of Alaska's tidy Eskimo village called Nome, Alaska, they couldn't get the antitoxin. The closest antitoxin was in Anchorage to Nome because the sea was frozen over the winter. The engines would freeze. The airplanes could not fly because they didn't have controlled temperature engines at that time. The only way they could get this antitoxin from Anchorage to Nome was with twenty relay dogsled teams. Balto was the lead dog in the last team that brought the antitoxin from Anchorage into Nome, Alaska and saved the town of over 2,000 people.
The reason we chose Balto is, in that story, Balto is using his sense of smell and doggy abilities to corral all the dogs so that the antitoxin gets to its destination. Balto didn't say, "Let us go to Nome. I want to go save all these people." The human, the musher, the person on the back of the sled is in charge. We think it is the relationship that people and AI need to be thinking about building in this next era, where there is AI that has abilities that we do not have, like infinite memories essentially and they do calculations in an instant. Those are not human things but there are things that are particularly unique experiments. How can we make sure that people are in charge and deciding the destination but AI is working alongside just like humans and dogs to help them get there? That was how we ended up coming up with the name.
Marc, this has been a great conversation, just like the first one was. I appreciate you joining the show. If people want to get a hold of Balto, how would they reach you?
Thank you, everybody. We will catch you on the next episode.
About Marc Bernstein
Building the market leader in real-time guidance for contact centers.
I'm the CEO of Balto, a tech company that helps contact centers close more revenue from their phone calls than humanly possible.
|| http://balto.ai || @marcjbernstein on twitter
I'm also the host of Reimagining the Contact Center, a warm and conversational podcast that digs into the new contact center economy and how technology, businesses, and the modern customer work together to build better lives for ourselves and our families.