021: How customer data is flipping the business value of the contact center
20 minAudio
PolyAI's Global VP of Solution Consulting Brad talks with CEO Nikola about the future of customer data in the contact center. They discuss the value of data in improving customer service, the challenges in gathering practical data faced by traditional contact centers, the role of AI in capturing and analyzing data, and the potential benefits of using data to inform business decisions. Follow PolyAI on LinkedIn Watch this and other episodes of the Deep Learning pod on YouTube
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
BradI think a really good concept to have is if you've ever been a meeting and we discussed these reports, we play these back to the customers. There's a big difference between saying I think versus I know and data gives you that kind of power.
NikolaHi everyone, and welcome to the PolyAI podcast. Today I've got a special guest, Brad Stein is our VP of Solutions Consulting. Brad joined us about four years ago now, Brad, right?
BradYeah, four years will be five in November, long time.
NikolaMy God, we've aged together. But we were about a small startup playing with machine learning. And I can say with no doubt that Brad taught me. Most of the things I know about telephony and the context center. And in this episode we'd like to talk about something that Brad is really passionate about and something that, we've been doing a lot of and monetizing in a good way. And that's really the nature of data in the context center, right? So people think about our product and about automation in the context center as something that, brings value because it improves customer service and cuts costs.
NikolaBut really. Once you get into having a really good conversation with the client, either between two humans or between a voice assistant and a human, a lot of good data comes out. So Brad, do you want to tell us a bit about that data? why is it valuable? What tangible impact to a contact center or a business does it actually have?
BradYeah, sure thing. So I think now what you have within contact centers, especially with conversational AI systems, you're moving away from single utterance type of intent. I have a problem, help me. Track and parcel, whatever that might be. You're getting a lot more context. And so what that means is that when you have more verbose utterances, you're able to collect a lot more insight, a lot more meaning about what the customer actually wants to do. And so you can actually transform your contact center into more of an insight center, where if you actually think about it, people don't really want to get in touch with businesses these days. They want to be self served.
BradThey don't even really want to call. So if you're able to pick out why people are calling, understand those problems more deeply, you're able to just fix those problems or try and attempt to fix them before they even happen.
NikolaThat sounds nice, but what does this mean in practice? What's an example of this?
BradSo a really good example would be, Let's take an example in terms of logistics. Hi, I'd like to track my parcel, right? That's a pretty simple intent, but what if it's like, Hi, look, your parcel, my parcel was supposed to show up yesterday, I'm getting married tomorrow, where the heck is my wedding suit? I really need to have something happen now. Standard types of systems.
NikolaIs this real story?
BradNo, it's not. Touch on wood, but I'm sure it happens. I'm sure it happens. And the problem is we don't know how much is happening these days because there's a lot of systems that don't track that. I think that there's a couple of different things that come out of that. So you would potentially create and flag for potential management escalation to see if there's a special type of insert that can happen. You're also trying to track and trend, how many potential other deliveries are impacted by that potential problem that's caused that delivery to be delayed. So is it, let's say, for example, Oh, A driver issue. Is it a depo issue? Is it because of plane was delayed or can't take off for any particular reason and start doing more proactive support to try and outreach those people so that, you're providing better support, but then also trying to figure out how do we prevent these problems happening
Nikolafor today, right? Let's forget about even an AI system and going to a regular contact center like this kind of management information. How do people get it right now? What does it look like? Is it good?
BradGood question. So it's usually post call. Right. So it's post call analytics where they have to run expensive processes programs to try and understand why people are calling. So customer data today is really obtained from two types of areas. It's either directly from the agent, post call where agents have to enter and wrap up information. But the problem is, Contact centers, especially very mature contact centers can have over 250 different reason codes or wrap up codes. Or it's even a product of somebody writing in unstructured text, which just doesn't really provide a unique aspect of why people are calling. And it's also really taxing for the agent, especially new, newly trained agent.
BradThey don't have, they don't necessarily know what to put in there or it's just causing kind of time lag. So it's not really efficient. And then the other way people typically do is post call analytics, where you're having to mine the data, look for particular keywords, but that always has to be hypothesis driven. It's not really leveraging the data to tell you or give you insight into what's trending or happening at any given time.
NikolaOkay. So then like in a world where A voice assistant is doing this, what we always thought about is flagging to people, something that they didn't know about automation opportunities, but also things that may be happening that they can, as you mentioned, deal with so that the call doesn't happen to begin with. Right?
BradYeah, it's also another piece of what's really interesting about contact center data. Now, if you think about everybody's working from home. Old school contact center. You'd have everybody in the office. You'd have a team leader in the pod. They start to hear and pick up on key keywords. So some of the things that we've been helping our customers as they go through these innovation programs is to help track and trend these keywords for a really good example. One of our customers, they're a large cruise line. They implemented a new piece of technology on the website and started receiving a lot more inbound calls for this particular problem. We were able to flag this up. And react to it and provide add additional intense responses to be able to handle this where, in a working home from environment, this person may get that once or twice, whereas we're able to see it holistically across the group and kind of react in real time
Nikolaas well. Absolutely. So then when you think about like the hive mind, then we're harnessing every single conversation happening at a time, reporting it back and turning the contact center into a real time analytics tool for the business, right?
BradYeah. And the thing is, is that what this will help do going back to logistics kind of areas is you can also start to track, particular reasons why people are calling, let's say, from a product level. So is it a product or service that's causing a problem? Or is it a particular region that's causing a problem? Or is it a particular language that's causing the problem and identifying what's happening there? So what we're using is some of this data, not only to just improve, let's say, online self service, but to prove operations to say, hey, look, These three or four depots have a particular problem. They're out of trend. They're having more calls for this particular reason.
BradThere's a problem here and you should investigate that. So we're also flagging up for CX leaders, how they can work with their operations teams to basically enhance their service or enhance what they're building to prevent calls in the long run.
NikolaSo then that basically makes the contact center more valuable to operations. Maybe even to marketing who like upstream revenue functions or
Bradyeah, I mean it makes in my mind It makes it more valuable for the entire business. So traditionally people view the contact center as a cost center ceos hate to pay for it it's usually a little bit under loved from a technology standpoint, but what this type of technology does how can I help? It enables you to collect what's good, what's bad. People are, millions of people are telling you every single day, what's good, what's bad, what they want, how you can improve. So if you don't use that to your advantage, you're leaving a lot of kind of value on the table and with this type of technology, we're helping CX leaders basically, become an integral part of the business and not just being viewed as a cost center.
NikolaThat's really, really interesting. And then when you think about. The future of this, capturing this data, analyzing it, where do you see investments coming in? how will people get this data? Is it from analyzing like human to human? There are many companies doing this already, right? Outside of automation, just looking at sentiment and like what I've always found interesting is, I've actually always found it a bit wishy washy, when they say, Hey, we'll track sentiment across, these conversations between humans, and we will give you insights out of it. It felt a bit. Off because, maybe you can get the general feeling for how a contact center is doing, right?
NikolaBut like actionable insights are very use case specific, right? And it really depends on what the conversation is about and what is even within the realm of possible for that contact center agent, right? In terms of the tools in front of them. So like when we think of the future for this, what are the investments people need to make? What's important, right? Is it worth doing?
BradYeah. I think the most important piece is, a fundamental understanding of what is possible with the technology, but then also what the customer is, what the customer's business is doing and, where they're, where their goals are, it's going, right? The first piece. And then what that really means is once you can understand the business, it's about combining the data. It's about combining our contact center data that we're pulling from customers. Why? How people are calling when they're calling trends, analytics and matching that along to the business processes. Do we? Can we? What if you could tell? Hey, every Tuesday?
BradOn the fourth month of the year, I know I'm going to get X amount of calls and I know they're going to be calling me about XYZ. How can I better prepare for that? And this is when you can start to build up. Once you run this for a couple of years, you can start to segment customers and profile customers. So Hey, this is John. He's 22 years old. We know he's going to call about this. He wants to talk to us about XYZ. He also would like this piece to be added to our service, whatever this kind of fictional company is offering, right? Why don't I feed that back to the product team? Why don't I feed that back to the marketing team? And understand then from a deeper level, which segments are growing, which are declining.
BradSo I basically become extraordinarily efficient as a business ensuring that marketing is tied up with customer service. To provide, the best products and services that are out there on the market in whatever industry or whatever product or service that you're offering. And that's the future, right? Where you marry each part of the business together, significantly break down the silos of Hey, we're selling this product and people don't like it. This is why they don't like it. Product team fix it. And so if you can start fixing those problems, you'll start to see contact rates go down. Become more competitive in the market. But a lot of companies are just starting
Nikolawhat I'm hearing. What I'm hearing from you is the contact center as it is right now is they're holding the battle line, right? It's under heavy pressure. It's under attack on 20 fronts and it's For what we see for the most part, barely staying alive, right? And what we all know the churn is high, people are leaving, it's hard to rehire, it's hard to train. What you're saying is hey let's quantify, what the problem is. Let's see where we can eliminate the problem at source, and your contact center just the visibility of everything so that it stops just reacting.
NikolaAnd starts informing the business as a whole. So does that mean that really, your CX leader is. In a way, perfectly poised to be, the operations leader of that company then because all of a sudden it stops being this black hole in which you're pouring money just to plug the hole of your product. Because at the end of the day, customer service is people mostly calling you. They're pretty unhappy. They have a problem with something you did. You're trying to be nice and pick up the call, but really it's something you probably did or a product that you didn't design. Well. That's causing the quote to again, right?
BradWe work in sales, right? I never take an opportunity not to speak. I always want to speak with my customers. I want to hear their feedback. In the market today, we see endless marketing emails. We see endless surveys, constant requests for feedback. But when it comes to the contact center, people like you. I want to put them on chat. I don't really want to talk to him about the problems. And that's such a bad philosophy to have. You should have the view of, I want to talk with my customers as much as humanly possible on any channel that they want. I want to be connected as much as possible. But you need to automate that, right? So it's just this is very, like the entry point. So I think when people are considering conversational AI, yes, you have to put out the fires of, SLAs, but where this is going, once you reach run state, It's not run state.
BradYou're constantly evolving, constant reacting to what's happening. And, a lot of consulting types of, let's say, efforts go into understanding the data because, having these massive data lakes, people talk about data lakes, oceans, I don't know, but all types of bodies of water with tons and tons of data. But I haven't seen anybody leveraging this in a really kind of concise way. And it's well, you need experts in each type of realm. And, we marry that with our customers where we come in and say, Hey, look, we know the trends, we know what people are saying. How does this impact your business? What's happening? So that back and forth is where, businesses are really going to start to overtake competition wise, there's a really great Gartner quote.
BradThey said, like in 2025, people are going to be competing on customer experience. If a product is non differentiated, people are competing on customer experience. I don't think that's really such a true statement because, know, with AI now, everybody should be offering excellent customer experience. I think where they'll compete is the data that they receive from the interaction with their customers and the ability to turn very quickly and move forward with those types of insights.
NikolaYeah, I think we're jumping a bit ahead, but you're right. Once you have a good voice assistant, it is there to completely alleviate the pressure of the contact center so that those guys can then do meaningful work on the product, the company itself. We've been too many. I, one comes to my mind where people are literally on Slack sending messages around ops and product teams because they have the time to actually do that. Through just like being a premium product with high customer experience, this was a contact center where people are using MacBooks.
NikolaThey've mostly been there for a few years. They know the product really well. They know who to reach out to in other departments to tell them Hey, like this thing's happening again. I think this particular thing was a query about adding a debit card to an Apple wallet. That's one of the probably the number one individual call reason to any banking contact center, because you just have to add it directly. It's and the
BradiPhone stolen, lost all the cards, but Hey,
NikolaI don't know. But it's really interesting. Cause then what you're saying is like, when we go a bit further, it's really like, how is the company using that data, that sieve and one, one big thing is like jobs change with AI and a lot of people worry about, like everyone's jobs going away because. These things are improving at an alarming rate, but I guess the truth is when we start gathering this kind of data, someone who has a bit of like breathing room there is looking at what's happening and trying to think about it. And what we found, I think, across many of the verticals we work with is that they come up with different things.
NikolaFor example, when we started working with restaurant groups, I thought we were selling them the convenience of not picking up the phone. And that's why we started, right? But if you remember, we figured out that for many restaurant groups they're saying no to large groups and Saturday evenings, mostly because site managers don't really want to deal with, groups of rowdy people having five pints each. And then, the whole system is configured in a way that just gives them a no rather than actually accommodating them because they're big moneymakers. I think this led to a revenue increase of about one to 2%, which to me was Completely mind boggling because I didn't know that it would be within, our abilities to give that kind of ROI.
NikolaYeah. With a voice assistant.
BradNo I completely agree. I, I remember, when we were looking at those particular reports and we found out, again, going back into the data, some of those sites were actually shutting off their online booking system. So we weren't able to make reservations. We just found out. There was an issue with that particular site. So we're able to flag this up kind of a real time. I think a really good concept to have is if you've ever been a meeting and we discussed these reports, we play these back to the customers. There's a big difference between saying I think versus I know and data gives you that kind of power. How many meetings do you have to sit there and wait to run a report? Wait to get the analytics.
BradWhereas if you can sit there in a management meeting or a leadership meeting, you pull up a report, you pull up the dashboards and this has real tangible information and insights and trends you move away from this. Oh yeah, this is my gut feeling on that. Or versus Oh, look, this is exactly what happened. This, these are what people are talking about. This is where we did well, this is where we need to improve. And now you have your sprint already created for the next week. It makes management a lot easier. And it makes, these meetings go a lot smoother. I think with competing opinions versus relying, having a data lit decision.
NikolaAnd how do people right now? So this is like what we're implementing and, Some of our customers are doing this incredibly well, go back to a traditional contact center like you're leading it You're hiring people you're looking at churn your service levels are wherever they are that week And it's people are usually firefighting and I've met very few contact center leaders who are not like perpetually stressed but What are they using to look at these problems now? What information do they have right now?
BradYeah. And this is a big problem in contact centers, I would say today, because you're in a leadership meeting. Imagine your SLAs are down. Your service level is increasing.
NikolaFor the audience, those are service levels, for example. Service levels. So your answer rate,
Bradyour answer rate has decreased. Your abandonments and the amount of phone calls that you drop, they increase, your wait time's going up and you go, what is happening in my call center? And so you look at a standard report and you say, We're getting a lot more calls. Why? And so in that meeting, you then kick off a piece of work to say let's get the data team in. Let's do some call analytics and let's figure out why by that point, that piece of work is going to take two weeks. It's already too late. The data that you are then interpreting
Nikolalistening people coming in to do this, like one off analysis. Those are then presumably who are supposed to be taking the calls you're taking Like the pool of people out of like the battle lines and you're having them audit what the others are doing. Is that how it works?
BradYeah. Yeah. It's because you're any type of analytics program, you're going to need some you're going to need people there who know why people are calling to help guide the analytics. You're then running on data that's two weeks old. So your call volume could have dropped or it could have increased even more, but that call volume could have changed for another particular reason. So you're always about a two week lag time. So then you start to result in, guess and check. What if we try that? What if we try this? Could we do that? And then you try and see what impact that has on
NikolaBut that's not a valid A B test because you're running on the next week's
BradExactly. So you're in that, I think world, right? Where, with AI now and you have, how can I help you? And people are telling you exactly what's wrong and we can show you these trends. You're in that meeting. You're like, Oh our delivery time has increased by seven hours. We've missed, 42, 000 time slots last week. This is why call volume is up. This is why that happened because, there was a depo problem. It's an unpredictable type of thing. So we now know why the call volume is up versus, endless kind of tracking. You wouldn't necessarily have that in a traditional call center, call center world.
NikolaOkay, cool. Awesome. And look the last thing I wanted to get is your take on, we're using generative AI in a lot of our products and Our CEO, Ian, has the sentence, which is like information in an organization flows upwards in sentences rather than graphs, right? So I think with a lot of this data and especially like the management information and contact center, you look at the charts and like the evolution of it, and it's fascinating. You can look at it for days, right? But then in terms of extracting that one core insight, it's Right? That's what our smart analyst product is about.
NikolaAnd we've used it like pretty well to just query data and get an answer. But like, where do you see this going? Which role is this? Is this like a new role in the contact center that exists to like drive automation and feed? It's almost like a PM role inside the contact center, or is it something else?
BradNo, I think like CX leaders, contact center managers have had this want, this capacity, this need for ages, right? As, as call volumes have been, when call volumes are lower, people were easy, able to easier, able to react. They're able to listen to more calls. The house is not on fire for lack of a better term. But I think for now, if we're providing tools that make it easy for people to understand, I think that's the biggest piece. Like, when you sit there and you look and say, Hey, What is this, code or this reason code or this wrap up code? Why does it actually mean this? You have to ask those questions to your point about the reports.
BradIf you have to ask 20 questions, it's not easily understandable. So I think, from an A. I. Perspective, making reports easy for contact center managed to react on the role is there. There's a possibility now with a new tool for, I think, people to do what they've always wanted to do. It's just It's time to start the journey and start leveraging it.
NikolaWell, guys, if if you'd like to see more and hear more, Brad would love to speak with you. So get in touch with us. But Brad, it's been a pleasure and thanks for being guests in the deep learning podcast.
Three years ago, Simplyhealth's PolyAI deployment was only resolving simple FAQs over the phone. Today, they run voice, in-app, email, chat, and WhatsApp conversations on the platform. Nikola Mrkšić talks with Paul...
An agent that gets better every day compounds into a very different agent within months. That's the idea behind Wren, PolyAI's new release: an agent whose job is to make your other agents better. Some of what it finds...
Companies are pouring money into AI, but not everyone is getting the returns. Ian Jacobs, VP and Lead Analyst at Opus Research, thinks he knows why. In this episode, your host Nikola Mrkšić sits down with Ian to work...
When PolyAI built its Builder MCP, one client helped shape it: Vixxo Facility Solutions, one of the largest facilities providers in the US. In this episode, guest host Damian Sasso talks with Vixxo's CTO Derek Neighbors...
PolyAI just launched Dialog-RSN-1, its first audio-native model, and your host Nikola Mrkšić sat down with its builder, Matt Henderson, to unpack why it’s a game-changer for building voice agents. Most voice AI either...