Home > Columns > Executive Interviews

Balto Executive Interview

Click image below to view the flipbook

 

In her continuing series of live interviews with CX/contact center technology providers at Enterprise Connect, CrmXchange Managing Partner, Sheri Greenhaus, sat down with Balto CMO, Ben Lazar, to discuss how their AI solution can show agents the best things to say, automatically score 100% of calls, and alert managers for coaching moments in real-time. 

Where to you see customer service today? 

Many contact centers have failures which cause significant economic impact on the organizations. When customer satisfaction (CSAT) goes down, additional products are not sold or they losing customers on a reoccurring type of subscription because customers are not satisfied. Consumers have very high expectations. The call center, unfortunately is often not realizing the desires of the consumer. It's really simple. Balto’s mission is to help call centers overcome the problem of huge amounts of turnover. We recently had a prospect with over 300% turnover …… and that number is not uncommon. This means companies have a revolving door.

Balto has come up with a number of solutions to help call centers reduce the time it takes to onboard an agent from two months to two or three days.  We help agents say the right things that are circumstantial … and those circumstantial things change very often based on what the customer says to the agent, it’s not what the agent needs to say to the customer.  That's how Balto fundamentally works. The other thing is helping the agent understand a very complex environment with complex products.  The biggest problem agents have is they just don't remember what to say.

Balto basically solves that problem by helping the agent say the right things, at the right time to the customer, integrated with their CCaaS. So as an example, if you are using RingCentral it's literally a one-hour phone call to get the actual direct API integration. We're working through all the major CCaaS with over 50 integrations, direct API integrations. It’s essential in today's environment. If you're going to be part of the enterprise atmosphere, you have to be directly integrated and work extremely well. 

Two other key functions is we help them coach more effectively. Post-call coaching is very ineffective because you only get a small component of it. Balto helps you provide listening for all calls and coaching helps you grade and score those calls. It helps identify agents who need help or need additional assistance. The system also helps identify when a customer's asking for an escalation so we can help diffuse the escalation. One of our customers have been able to reduce call escalations by 50%, that’s pretty impressive stuff. 

On the QA side and calls post-call is after the fact you're getting to an agent, you're talking to them about a mistake they made.  QA and compliance can be huge. If the agent does not read a disclosure statement, they can get in a lot of trouble. And then if you have to go back and do discovery, doing manual call discovery is extremely difficult and inaccurate. We have a way that that can be done with a great sense of accuracy in a small period of time. So you can identify any potential compliance or QA problems remediate them before no potential lawsuit takes place. 

So, this is happening in real time?

It is real-time guidance. As you are speaking to the agent, each system is telling you basically essentially what you should be saying next to the agent to overcome objections; like the price is too high or I'm unavailable, or I can't make my payment.

Is this whispering? 

The coaching can be live whispering. If a coaching situation rises, they're immediately notified on your desktop… the manager….and they can kind of whisper mode directly to the agent.

You can basically have sentiment such as if the customer’s agitated.  There's a number of things that can take place inside of the technology, because the technology utilizes the combination of machine learning and natural language processing. Natural language processing understands that human and tech and can kind of basically parse the language to identify what to say next.  The machine learning component basically utilizes all of the database access to pull information that the agent may not have access to. Instead of flipping books, shuffled through papers, they can use Balta to identify very critical information that they need right on screen.

Does Balto have an API into their CRM system to get all that information? 

Whatever’s in CCaaS system is accessible.

If the agents are having a hard time and they need to get some information, is Balto listening and brings it in or they have to type the text? 

It's listening to both ends of the conversation.

If I'm an agent and I need to know if this dress comes in red, the system is listening pops in the information, so they don't have to click?

It brings that information. If they have to go deeper, it’ll bring them a link to click through to, but typically it's going to get the information, and that call is logged and tracked. If needed, they can go back to it for discovery or training purposes.

So now with machine learning if the agent came up with an idea that nobody thought of before, but it was perfect, does that become accessible to others? 

The information would be provided to the managers. Agents can't change the playbooks, but in our future iterations, we're looking at creating agent driven playbooks.

And then never missed a coachable moment. The coachable challenge is when you take one to 2% of your calls, which is about the average, and you're looking at that specifically, and then coaching basically off of that, it's a very small window. It probably is going to give you a false positive, but with Balto, you can see scoring all of their calls. They're doing hundreds of calls. You can see and drive down into them and see where they failed and succeeded in those calls. That's not based on one to 2%, it's based on 100% of their activities.

Balto is scoring without any human intervention?

Correct. And we can modify those parameters based on the requirements of the individual customer. They may call, they may score more heavily on things like call conversion into a sale or call conversion into an appointment. Those types of things, we can modify the scoring.

Is it looking at compliance … they followed everything correctly? 

If you miss reading a disclosure statement, you're going to have one, you're going to have a QA issue and you're going to have a detrimental call score.

We talked about scoring calls which we do that automatically. It's pretty impressive. You can sort search. It's got all the analytics, all the reporting. It can tell you which agents are at the top stack rank. You can publish this. A lot of our customers use this as a way to incentivize and gamify their agents. So, if you're in the top X percent or the top agent, they'll give you a prize or some type of monetary benefit. If you're on the bottom rung, they can identify very quickly how to coach you, where you're specifically coachable and what they need to emphasize. They then have a history of all that coaching. They can see the person who is improving or not, and they can do this at scale. With the old methodology a manager who runs a group of people and it is a highly subjective method of assessing people. Balto provides a very clear and specific, singular way of managing all the people; not only your agents, but many of your managers as well.  Companies can look at who well the managers are doing with their agents and if they're not progressing their agents, the managers can be coached by their executives.

What reports do the executives get? 

The simple reporting of these agents and you can basically see the relative change in over time. You know, what who is  improving or not improving, do they have the lowest or highest and agents that type of stuff, it's just basically gives you really good historical graphical information or reporting and succession or analytics.

The interesting thing is that you can see results.  UGA spoke on our behalf last week. They actually were seeing these results within 30 days, to be able to transition as all escalations, to increase collection rates, and increase customer retention rates.

Are there reductions in supervisor requests, because it's telling them what the supervisor would tell the agent? 

Imagine you were being coached after a game through game film but now you're being coached actually in real time. You come in through whisper mode or they can come in and chat with them and basically say these things diffuse the situation. This happens very quickly because they were able to give immediate biofeedback to their agents. Our primary use case is selling collections. Outbound selling or inbound for retention.

 

I can see inbound for either financial products or something similar. So on the slide ExpoHome shows 162% increases? 

That's the customer's data.

Because they told them what to say and how to close it. What do they attribute that to? 

People are not all naturally good influencers in selling and some agents are good at resolving customer issues, but they're not good at selling naturally. So, when you get an agent and you have a very specific variable script that tells them what to say to a customer based on their response.

I guess if there's a lot of products, it reminds them what to say. 

That is the number one. We did a survey of 2000 agents. And the number one gap was just for that. We're doing research right now with Harvard Business Review.  We have over a hundred million log calls almost probably 25 million now. We have huge amounts of data that we can do amazing analysis on. Educational institutions have the academic knowledge, they don't have any of the data. We're partnering with a number of large educational Academics.

So now go back to when you started. What's the origin of all of this? 

Our founder, Marc Bernstein, was an actual inside sales rep agent. He was one of the best I can remember. His boss was managing and giving him call coaching and Marc would forget. He built a spreadsheet with macros that told him what to say and he typed it in to a spreadsheet and it went through list and pulled up the next thing for him to say. He essentially built a very crude example of Balto in an Excel spreadsheet. And then our co-founder, our COO said, we've got to make a company out of this; they built Balto. 

What is Balto’s sweet spot? 

Our primary sweet spot are enterprise customers with 500 plus seats. We do work a lot with companies that are between 50 and 500 seats, but obviously this type of value really showed very well scaled and the technology can scale and it's a SaaS solution. There's no localized requirement of installation.

Do you have agent feedback? 

We get agent feedback all day long through NPS scoring. We get 20 to 30 NPS scores a day, maybe 50.  Right now, our promoter score is 27 and we have 57% of our agents are net promoters, which means they're a nine or a 10 out of 10. And our goal is to have that to 70%.

So what's going to make you get to that percentage? 

We're continuing to make some pretty important improvements in the products. All AI has some natural language processing error rates. We have lower rates than Google's AI, on natural language processing. As we improve that it will help us. We're also making our playbook and scripting more personalized to the agent.

Within certain parameters each agent will be able to customize their own stuff. Right now, it's only manager scripted. If an agent speaks a little bit differently or has a better methodology, they will within the limitations, have that ability. We think that that flexibility will give us better NPS.

What other languages do you work in? 

We are presently working on Spanish. It is a big build because Spanish has so many dialogues. Mexican Spanish versus the South American Spanish versus a Caribbean Spanish are vastly different. There's some complexity to it. It's not easy.

I would think it is. So when do you think that would be ready? 

We'll probably have Spanish this year, late third quarter.

What other types of things are you looking at? 

Chat. What I would characterize as an assisted chat where you're chatting with a live person, not a chatbot. Think about someone who's trying to type and trying to access information. How many sessions were they managing versus somebody who was actually given recommendations by Balta to say the right thing while they are chatting.

What else would you like our audience to know? 

Balto is directly integrated with over 50 CCaaS systems, which immediately drives adoption and we can get them up and running very quickly. The other big item is taking a change management approach with our clients. We have a dedicated customer service success manager that basically meets with them constantly is driving all of the reporting to make sure that their metrics, the conversion rates are continually improving; versus a lot of technologies that come in and they drop it in and they leave.  We're part of the technology implementation. The easiest part if the implementation.

How long does it take to get up and running? 

I would say 30 days. We have that conversation, excellence lab. We’ve created the best practices by industry or playbooks we come in with preset playbooks and we can edit these and get you running quickly. We can get that integration done in the day. We can get the people trained and we'll get your primary playbooks within a week.