Greg Kilstrom (3:58)

Great, great. And so your position, your title here you lead the PEGA decisioning platform team within Verizon business. For those a little less familiar, I'm very familiar with Pega with a few clients. But for those a little less familiar, could you describe what is that platform and at a high level, how are you using it?

Kunal Mangal (4:20)

Okay, yeah, sure. I mean Pega is a big software company. They've been around for I think more than four or five decades probably. So they create, they have things like CRM and business process automation softwares, but they also have this thing called what we use is called Customer Decision Hub. What it is, it's a portal that allows you to implement automated decisioning strategies. It's a fancy name for maybe recommendations. You can say so what it does. So the main advantage here is that there's this one portal. You can bring in data from various sources. You can create business rules, you can create machine learning models. It has built in experimentation abilities and all that. So you can manage all that and plus your actions and stuff, your recommendations, what kind of things you want to recommend, those things all together you can manage in a central sort of low core type of environment and then you could serve it into different channels of engagement platform just using standard API architecture. So in a way, I mean you basically use one brain to figure out how to decide on if somebody walks in, what's the best, how to, what's their need, what to recommend for them. You can just decide all that in one place and you can serve it out to different channels of engagement. So if you do it well, you're giving a real Omnichannel experience and then you're using feedback from what happened to my recommendations from multiple channels to further enrich your AI models. So that's what it is. In essence, I would say there's nothing you can do with it that you cannot do without it. It just makes it managing easier at scale. And I think the go to market is faster because there's a low code environment, as I said. So it doesn't take a lot of time to build new strategy or bring in new actions and stuff like that.

Greg Kilstrom (6:10)

Yeah, because I mean, part of the challenge with a lot of this stuff is just getting the right pieces aligned. Right. Because it's not just platforms, it's teams, it's data, it's all of those things. So just kind of centralizing that in one place can be really helpful.

Kunal Mangal (6:24)

Yeah. And plus another thing is that, I mean in organizations like you, like Verizon, we have so many channels of engagement with the customers and if you want to implement the same logic or same sort of intelligence, multiple, you have to do it multiple times. So better do it in one place. So yeah, that's. I mean, but you're right. I mean the centralization in this case helps. I mean, although I know that it's not good for everything. So that's the idea. But it's still a challenge to kind of do it. Right.

Greg Kilstrom (6:54)

Yeah. So you mentioned in your experience you've worked with in a number of different industries. You've worked on the B2C side as well. What's now working in B2B? What's been your experience? You know, what's, what's been different from your experience? And you know, what are you seeing there?

Kunal Mangal (7:13)

Yeah, I mean, I started in B2C and to be honest, I'm not like a hardcore nuts and bolts marketer, but I kind of learn along the way. I think there some ideas and some concepts are very much similar in B2C and B2B. So for example, we talk about like there was a session earlier, we were talking about personalization. Personalization is equally important. B2B and B2C. Now it worked in different ways. I mean, but basically you're saying tailoring messages and offers and experiences based on data driven insights. It gets more sophisticated actually in B2B. Because think about it, you're working with multiple stakeholders within the same client organization. If I'm talking to the CTO of a company and try to pitch something versus I'm talking to the HR head, their needs are different, their perceptions are different. So how do you personalize your context with the customer. I think that if you learn some techniques or best practices in B2C a lot of them are equally applicable in B2B. So that's one thing. Same thing with customer experience, I would say. And you see actually smooth, hassle free experience from initial contact to making the sale, onboarding, ongoing support after that it applies same in both business customers also they want hassle free experience. So again, if you learn something in your B2C, you have to understand that's equally interesting and important. Some of the differences I would say is content marketing, although I mean it is there in B2C as well. But I think it's more, it gets much more sophisticated in B2B because the nurturing period is much longer in B2B. You know, you're looking at like, I'm sorry, I'm speaking from a telecom perspective probably, but you know, I mean, new model comes in, everybody wants it, they're just looking for the best deal. But in B2B, you know, you don't sell stuff by exciting people. You have to show them the value, you have to understand their pain points. I think the nurturing part, the content marketing part in B2B is something that you touch upon in B2C. But I think it gets much more sophisticated and complex in B2B and when they talk about things like Gen AI and all that. So I see a lot of that, a lot of people trying to figure that out.

Greg Kilstrom (9:34)

Yeah. And so a lot more time spent in the journey and a lot more potentially decisions and kind of hurdles to get through in the process.

Kunal Mangal (9:42)

Definitely. Yeah. So that's one thing I'm learning. I never took it seriously in B2C and I was like, whoa, it's a big deal now.

Greg Kilstrom (9:49)

Yeah. So a big part of your role involves data driven decision making as you said. And you know, a lot of that has to do with data, a lot of that has to do with platforms. But there's a huge people and process component to that as well. Right. So you know, from your perspective, even having the right tools in place and the data connected and all of that stuff, what's the mindset shift from a. Well, we did it this way and it worked. That anecdotal perspective versus a really, truly relying on data to make decisions.

Kunal Mangal (10:25)

Yeah, that's a great question because believe it or not, that is the biggest challenge. So I think the mindset change has to first of all start from the top. You know, you have your topmost leadership needs to demonstrate that they are using Data for their own decision making. Now they need to be seen as doing it. You know, you go to meetings, they need to ask for data, insights, evidence. You know, you go for a proposal approval. You have to understand that my leaders are going to ask me for concrete evidence. So that is one of the first things so that everybody knows that, okay, you know, we, we value data and we value fact and reality. So that's one big change, I think. And many times these initiatives tend to. People tend to start them from like one division or one little corner of something. It doesn't work, it's not going to scale well. Second thing I would say is the culture of collaboration and the cross functional cooperation. Because see what happens is if you're saying we're going to start using more data to make our decisions, first thing is you have to get that data and it doesn't stay in one place. Earlier if you run a business unit, you would say, I own my data. So you have to open your data up to other. Because you have to create some data layer which is not just your business unit or your department. So open that data up to other people. Open your decision making processes up too. And that could create insecurities that could also create. Because if you're not seeing the value like why are we doing all this? So that is I think a big mindset change also that you have to learn to collaborate better. I think third is probably having a culture of constant improvement, learning and curiosity. And this may sound preachy, but think about how data science works, right? I mean, you cannot have the most perfect model on day one. It needs feedback, it needs more and more data to constantly improve itself. Plus things change, customer behavior changes the way, I mean like new data sources emerge and all that. So if you don't do that, that culture of where folks are questioning things and saying, hey, we built this model last year, but is it still good? Or so basically are you measuring it constantly? Are you willing to work on continuous improvement? I think that is also that kind of culture is very important too. I mean like so employees should be encouraged to ask questions, to investigate. And you know, so I think there's a couple. I mean, so these things, I think from mindset perspective, you know, leadership demonstrating the need and then business units opening up and collaborating more and trying to incorporate that at an employee level. That culture of, you know, curiosity.

Greg Kilstrom (13:18)

And so from that really talks about from the internal perspective. What about from that customer perspective? I mean at the end of the day, everyone is driven towards sales and revenue growth and Lifetime value and stuff like that. But how does the customer experience kind of weigh into this notion of data driven decision making?

Kunal Mangal (13:42)

This CX remains the most important thing. I mean, honestly. So no matter how smart your decision making is, if you're not communicating it to your customer at the right time, in the right context, using the right channel, it's not going to work. And it has, I mean, it has happened to us many times. We thought our setup was great, but we didn't see much adoption or we didn't see much success. So one thing is, yeah, I mean, no matter how great your data driven insights are, you got to figure out from a customer experience perspective, where does it make sense the most, how to present it, what content should be delivered to convey this action or whatever to the customer. I mean, in fact, in company like Verizon, a lot of our efforts of spending or creating data driven initiatives and projects are geared towards improving customer experience. And you measure customer experience using multiple things. But they're saying, okay, well we have gaps here, Net promoter score, we want to increase, so how do we use. So a lot of actually effort is going towards improving cx.

Greg Kilstrom (14:51)

Yeah, yeah. So that's desired end result essentially is that cx. Yeah. So you talked about the feedback loop and the importance of that. There are a number of ways to get there. So you talked about how we're not going to get things perfect on day one. The idea of machine learning even and all those things is continuous improvement. What about another way of doing that is to do the pilot project and the small thing and grow from there. In an organization like yours, is it does one work and not the other? Do both work? Does it kind of just depend on the use case?

Kunal Mangal (15:36)

No, I think the same sort of iterative approach works better. And part of the reason is, first of all, no matter how much confidence you have in your ability to build the best models in the world, you have to understand that the outcomes from technology intervention projects or automation projects is much more certain. You can measure, okay, if I automate this process, I'm going to save 100 hours a week and blah, blah, blah. But when you are bringing in probabilistic reasoning, machine learning and all that stuff, then the outcome becomes a little less certain. You don't really know, honestly. I mean, you know to an extent, but it may not work out the way you expected. So you have to always start small, find a very specific problem which is not too complicated and try to solve that using some data driven approach. I'll give you an example for Example, when people call to disconnect reps were trying to save those customers and they were going to go to the deepest discount and say I want to save you. What if you give you 40% discount, save this line. So yeah, fine, your retention rate is good. But then your cost per offer is also very hi. So can you solve this simple problem? Maybe just try to gauge the customer's lifetime value or probability to churn and all that, whatever. And then if you can even gate those offers, maybe we start with 20% for this customer or things like that. And the way you design your interface, you make it work that way that nobody can jump or even that thing. A simple problem doesn't require a neural network model to solve, but you can show value. You can say two easy KPIs. What is the disconnect rate? What is the cost per offer? You prove it. That gives your stakeholders some confidence that okay, well it is working. And then based on the feedback loop and all, and you basically try to improve that process again, when a KPI start to drop, you figure out, okay, what do I do? So that's one thing. The other thing is now it has given you a case study, a success story. Now you can use that to basically figure out more use cases, go in front of modes. So this is the only way it's going to work, honestly. I mean if you're a small company or a startup, yeah, maybe you could do sort of a top down that way. But I think in companies, big companies, you have to do iterative because as I'm telling you, I mean like your value estimates are going to vary from reality, but you want that variance to be as small as possible because once your stakeholders lose faith in data, it's very difficult to get it back. So I mean from my perspective, that is the best way to go. Yeah.

Greg Kilstrom (18:31)

And another thing, you touch on, you touch on next best action which is okay, a customer is about to churn, let's send them some kind of reminder or whatever. But it's also next best offer, which is what you also kind of. So in other words, it's not just okay, it's time to like communicate with this person, it's what's the best way to retain what's the, you know, it could come down to the cost of a text message versus an email, you know, in some, you know, really broad use case or something like that. So it sounds like, you know, it's when done. Well, it's a nuanced thing, right?

Kunal Mangal (19:08)

Yeah, I mean, definitely. So you cannot Create unrealistic expectations. And that's why you don't want to start very big, where the value proposition gets a little bit vague. You got to give. That's how you're going to get executive support, by the way, because they're nervous, they're not experts in machine learning or data science or so they want to see evidence. And if you start small, you can give more concrete evidence and you can prove it quickly enough also, versus if you do a large project which require a lot of IT work, a lot of new architecture and all that. You know what happens when any project becomes technology project? There's going to be cost overruns, there are going to be delays. So you don't want to go that path right from day one.

Greg Kilstrom (19:53)

Yeah. So let's talk a little bit about. So your data driven decision making involves AI. You know, there's lots of flavors of AI. So you know, I think we could talk about a number of different ways that you're utilizing it. But I wanted to talk about generative AI in particular. You know, I'm familiar with some of the features in PEGA that utilize generative AI. But can you talk a little bit about, you know, where you are on that kind of adoption life cycle and you know, where do you see promise with generative AI?

Kunal Mangal (20:24)

Yeah, sure. I mean, first of all, I wish it was there when I was in college.

Greg Kilstrom (20:28)

Right, right.

Kunal Mangal (20:29)

My wife is a professor and she always complains that so many people are using it now. So generative AI is actually a much broader topic, at least like in my organization. So it's beyond what my team is doing. They're looking at it at a very big level, what could be done. So in my world and if I talk about PEGA for examp, so they are introducing some capabilities and I've looked at some of those. I mean, it's in the latest version. We haven't migrated to that yet. But some of the gen AI capabilities the tools are implementing, tools like PEGA are implementing are basically to make working with those tools easier and faster. So for example, like, you know, if I wanted to test some decisioning strategy earlier, I would go in and I try to create a Persona of like, because I have some Persona of customers in mind and I want to test this decisioning strategy against that Persona. I have to set up that Persona myself now. I think in the new version you can just type in plain English that gave me middle income, blah blah blah, low churn risk and they'll create it. So some of the features are like that the tools are implementing Genai. If you need a report, just write in English and our tool will give you a report and charts and everything. So part of it is that. But one interesting feature that I noticed is that again, I mean Genai from a marketing perspective is all about content creation. So some of the tools including Pega are now implementing some features and I've seen an example of it where you can create the, what they call treatment copy of the content automatically. So think of it. I want to show you an offer, right, but how should I display it to you? So currently we let marketing people do that. I mean we're not making that decision for them and say, okay, this is the best offer for Greg. Now I send that to front end channel and they apply whatever content because the content needs to be regulated and legal approved and all that as well if it's showing to end customer. But now using PEG app for example, they're saying you could just set up, you'd say, I want you to create that content for Greg for this particular offer and you can also choose like I think they have some sort of a shadinier 7 principle of persuasion kind of thing. And you say you want tone to be formal, informal, what principle of persuasion you want to use and then the genai would figure it all out and create. So I think the companies are looking at those kind of use cases. Another thing is chatbots, of course, the IVR service to sale transitions. So how do you interpret and how do you present the content in a way which would relate to what the customer is looking for and all that. So I think there are a bunch of stuff that they're looking at as simple as some repetitive tasks like when you write prospecting emails and we're trying to see how many hours each rep is spending per week for do that. And with Genai, can we reduce that? So that's pure efficiency play. So yeah, there are a lot of such use cases but everything is geared towards of course creating content. And I think another aspect is of course, how do you analyze all this unstructured type of data that you have in your organization, like conversations, emails, phone conversations, that's part of the technical side. But the output wise, that's where people like me are more interested in what kind of output I can generate.

Greg Kilstrom (24:06)

Yeah, yeah, but I mean to know that someone just had a bad customer service, I'm sure that never happens but you know, you know, someone had a bad customer service experience and then that feeds into okay, well you know, do we or don't we send them a marketing email tomorrow because they're probably still kind of annoyed at us. Let's, you know, let them cool off.

Kunal Mangal (24:29)

Exactly. Yeah. Like in telecom business, you know, network issues are very common. Right. I mean what, how do we know that it was. You really had a bad network experience which affected your business. Right. And if you could imply that with your conversation you had, or maybe you chatted with somebody or you had a service ticket. So how do we. Yeah, so I mean like there are a bunch of use cases. I think there is. They've created sort of some sort of a gen AI task force or something and they're looking at across the board. But we're looking at it both from a service sales, customer experience, loyalty, all perspectives.

Greg Kilstrom (25:08)

Well, so one more question for those out there that may be in a smaller organization or looking to become more data driven, what would a first recommendation to get started towards becoming more data driven?

Kunal Mangal (25:23)

Yeah, I think we kind of touched upon it. A couple more questions before but you I think start with something specific, a specific goal, what are you trying to solve and what does the vision look like once you solve that problem using analytics. So find something concrete because next step is you're going to go for executive approval and buy in. And again, as I said, they're not experts too. So they want to see some concrete evidence and some concrete value proposition. So don't try to boil the ocean and don't say I want to transform this whole company into a data driven machine. Yes, but you can't rebuild the machine in a day also because nobody wants disruptions. I can do a very good job if my boss doesn't give me any new projects for next six months because I can figure out a lot of things but he will not stop doing it. So that's one thing. You find something specific, solvable, do a lot of research, come up with a clear articulation of the value proposition, of the vision of what things will look like once you solve this problem, KPIs and all that, and then executive buy in and then you try to create a data driven culture which again I talked about encouraging people to collaborate more and think that way because data driven conclusions are going to challenge your intuition. Many times you thought X, but data says why, who's right? And if you're an SME, especially if you're in sales, they believe they know everything, which they probably do. But sometimes you know the direction but you don't know the magnitude, relative importance of two things. Right. So that you know, you try to create that culture among people. And again, how do you do it? Again, by showing evidence. If you picked up a good use case and you're showing people results, they'll start believing you. And finally, I think one thing that I realized is that you have to productionalize your AI or your whatever, data driven insights. It's not enough to have a room full of smart data scientists churning out models after models. You cannot tell a marketer that hey, there are all these propensity scores lying there. Grab that spreadsheet and use it. No, you have to figure out how to productionalize it in there in our B2B marketing processes automatically. You can't expect people to use the results of my analysis and figure out how I'm going to use this. You have to embed it in their processes. So that is very, very important. That gets overlooked many times.

Greg Kilstrom (28:02)

Great. All right, well, thank you so much everybody. Really appreciate.

Greg Kilstrom (28:14)

Thanks again for listening to the B2B Agility podcast. If you enjoyed the show, please take a minute to subscribe and leave us a rating so that others can find the show more easily. You can access more episodes of the show at www.b2bagility.com. That's b2bagility.com. While you're there, check out my series of best selling agile brand guides covering a wide variety of marketing technology topics. Or you can search for Greg Kilstrom on Amazon. Until next time, stay focused and stay agile.