John Abbatico is the founding CTO of Othot. Othot provides the nation’s most advanced analytics and artificial intelligence solution for colleges and universities. During this session, John explains the value of predictive analytics and just as importantly allays any fears predictive analytics will take away our jobs.
For the past few years, we have all be hearing and reading about the importance of Big Data. It is the fuel that propels us from traditional reporting, which is descriptive and can tell us what happened in the past, to predictive analytics, which can help us make inferences into the future. And the models get smarter as additional data is added (i.e. Big Data sets.) It is interesting to note that the human mind works a lot like predictive analytics models. The "BiG" difference being the models can you 1000s of data points.
It is important to note that predictive analytics does not tell us the future! People will still need to be involved in the interpretation. What is exciting is that the next step is what is described as Prescriptive Analytics. Prescriptive Analytics can tell us what will happen and then tell us what you can do to change the outcome! Coming back to Big Data. The more we know about the history the more accurate the predictions. That is why companies should have a strategy for what they are doing with all the data they collect.
So how does this apply to sales? Imagine what you can do with all the data being tracked about each prospect. Just like Othot helps universities determine who is most likely to enroll, predictive analytics in sales can help determine who is most likely to buy your solution. And the more data you are collecting about your prospects all the way through to closed won or closed loss, the more fuel for your models.
And consider the impact this has on resource allocation. People can apply their time to those opportunities and in turn become more productive.
John's message is clear - Go do this now - predictive and prescriptive will change the way we work. Get out ahead of this. Have a strategy. Start to understand. Collect the data. And lastly, if you have kids, encourage them to get into data science.
And don’t worry, one of the biggest myths - that Predictive Analytics will take over our jobs is not true. Predictive Analytics will be a lot like reporting. It will augment decisions. It is like saber analytics in baseball.
Connect:
Connect with John LinkedIn and learn more about Othot by visiting www.othot.com.
Connect with Greg on LinkedIn and check out Omedym’s website and their LinkedIn and Twitter
Greg Dickinson: Alright, welcome, everybody to the convenience economy, we talked about the convenience factor, and how, right everything in this is changing, and people are more prone to want to do things that are a lot less friction. And so, you know, some may be thinking that my guest today, how does those two things come together convenience and our topic today is predictive analytics. But I was fortunate enough to get John Abbatico, the CTO of Othot to join us today. And for full transparency. john and i were pre sales engineers together back in Ariba in the late 90s. And, you know, we did another startup together. And then we successfully exited our last startup together. So john, you know, is a good good friend of mine, and he's very really busy CTO of Othot and has graciously given, given this time today to talk about predictive analytics. John, before we get into that topic, give a brief introduction of john Abbatico and what Othot is all about.
John Abbatico: Yeah, thanks Greg. Great to be here. When I heard you do my introduction, I realized that pretty much half of my career was pretty engaged with you and, and and partnered with you. And I love the fact that you tell me, I'm your brother from another mother. And I think that's true. Anyways, let me tell you a little bit about me. Believe it or not, I started out as a nuclear physicist, and worked at a government lab for a period of time right out of college. Well, as a nuclear physicist, I realized that my real passion was in technology and computer science. And like many of us probably listening to this podcast, I did a career switch switch. And that brought me on a different track. Through my long period of technology, I've worked for companies like Oracle had a great stint at Oracle was a demo engineer at Oracle really loved that time. moved over to Ariba where I met, Greg did a number of things that at Ariba including heading up the technology demo team and, and doing demos, and even helping our technology path at Ariba. Left Ariba just do a startup with Greg in security space. So got a good introduction of how security, at least in the mid 2000s worked in in some of the importance there. That didn't quite work out the way we wanted it to. But you know, as you said, as they say you learn to fail. So you can learn to succeed. And that's what happened for both of us, I think Greg moved on to the SAP for a period of time, so awesome, had a chance to be at Oracle and SAP and see how the team and SAP operates when it comes to doing demos and making sure that the sales process runs as frictionless as possible. didn't stay there too long, because Greg and I got back together for our next startup. And that was hype heroes. And that was an awesome ride. That was in third party risk management. And Greg and I built that company from the ground up, I was a CTO, we built the technology. And Greg built the sales team. And we were able to, as Greg mentioned, have a nice exit from that company. And it's actually now owned by a publicly traded company named Koopa. So we're really proud of what we did there. One after that the exit though I kind of got the bug for the predictive analytics. You know, it, it definitely was a hot topic in 2014 2015, as it is now, I got the inspired to want to do something in predictive analytics. So I got together with some folks here in Pittsburgh, and we built a predictive analytics company called Othot, and Othots mission is to democratize predictive analytics. And I think we'll talk a little bit about predictive analytics. But that one is extremely powerful, but it's really hard. So our value prop was how do we deliver predictive analytics to folks that really don't have the resources or the know how to do it?
Greg Dickinson: Which is most of us so you know, john starts off with a with a nuclear physicist and and you know, I always joke with John, this is just not as hard as rocket science but then you know, when you when you when you have a hard problem, what do you do you go get a rocket scientist. And so john is very modest across his careers he's been solving some, some some pretty cool problems and and you know, CTO of ipiro, building that from the ground up and then now write predictive analytics and AI. And it's just really like he said, It's not easy, right? And I think what's really, that the audience is probably thinking right now convenience factor, right? How does that how does predictive analytics fit in there and so you know, we often think of convenience as Hey, I can I can now order my food online instead of having to call or Amazon brings me stuff into Two days I can stream instead of going to movie theater or, you know, but think about your job, right. And the convenience factor is in our work day, every day, right making it went making work easier, making it more convenient, less mistakes equal, less frustration, equals more convenience for you. And so, you know, john, before we talk about how you've taken this abstract concept of predictive analytics, and brought it into the higher ed space with Othot, let's start at the basics of, you know, for my audience of predictive analytics, so start as if I was in sixth grade, john, you know, what is predictive analytics, we hear a lot about it, right? A lot of people buzzwords, Ai, predictive analytics, all these things, and, you know, data lakes and all these big terms. But, you know, bring it down to the sixth grade level. My audience doesn't know what predictive analytics is, what is it? What does it mean? What does it do? Why would I Why would I care?
John Abbatico: Yeah, that's a great question Greg. And it it is confusing. In fact, when you think about AI and machine learning, it's a really broad area of technologies at play. And predictive analytics is one of them. So you know, when when you think about AI, and the computers taking over and facial recognition, all those things, they're all AI. But predictive analytics is a specific subset, as I said, and really, maybe to better understand predictive analytics, I a little bit of a history lesson, right? I'm sure your audience knows what reporting is. And I actually remember the data warehouse craze of the mid 90s, I was at Oracle, everybody needed reporting, you needed to have data and data was going to tell you everything you needed to know. And it it there was a lot of money and time put into building these data warehouses, which is pulling data across the enterprise. So you have it in one place, and then you can ask it a question. And then would give you the answer, well, didn't quite work out. There's the way that everybody had hoped to hoped it would. But it did start that the reporting error of our of our society, right data is available, it's digitized. Predictive Analytics is an extension of that. So I like to think of reporting is something called descriptive reporting can tell you what happened in the past, okay, it can't really, it can help you make inferences to what's going to happen in the future. But it doesn't really tell you what's going to happen in the future. Predictive Analytics is an extension of reporting in the fact that it now moves beyond what happened in the past, to give you real predictions of what's going to happen in the future using your data. So in our case, we use historic data about students and their behavior. And from that we're able to predict for a particular school year, what students may or may not attend a university, for instance. So it helps it helps give some insights into the future. Now it doesn't, it doesn't tell you what to do. So, you know, I think when you think about predictive analytics, it's more its decision support, telling you what the likely outcomes are going to be in the future. It's not necessarily telling you or doing it for you in an automated fashion, although it can. But as we evolved in predictive analytics, it's more of a here's what's likely to happen tomorrow. And what's really exciting is the next extension of predictive analytics is prescriptive analytics. I'm going to tell you what's gonna happen tomorrow. And then I'm going to tell you what you can do to change the outcome.
Greg Dickinson: Wow. So so let's bring that up. So what you know, because it's kind of kind of interesting, right? So what you what you said is that data in a data warehouse, right was what can theoretically said, Hey, I can now go ask the data a question, right? And theoretically, the report would be generated that would give me the answer to my question, but boy, I had in my mind No, right kind of all how all those things connected. So predictive analytics, you're saying, Hey, we can look at the report? And I think in your world, you're probably looking at, you know, did they take a foreign language? And did they do a college visit? And did they would they score in their SATs and you know, what, what size? You know, what size college high school? Did they go to I because I mean, all these quote unquote, data elements that would be in a report, if I ran a report of Greg Dickinson, here is his high school history. And then john, you're using that data then to say, Hey, is there a pattern, I guess, is what you're saying. Right?
John Abbatico: Yes. So that's the historic data. The data that most people are using is reporting of what happened yesterday, is actually the the training material for the predictive models. So exactly what you said, we have historic outcomes of what students did, whether they can't so let's take take a typical example that we do at Othot, we predict the likelihood of a student attending a university. And that may sound like well, what well, how is that valuable? It's actually infinitely valuable to most of the universities in the United States because they Most of them aren't the Yale's and Princeton's where everybody they accept comes, most of them are only one in four students that they accept come to the University. So now think about that, like, it's a classic sales problem as well. I have all this population, and I need to know where to focus my time, attention and money to make sure that I get the best students that come to the University. So how do we how do we train the model for that, we look at the past, we have all the outcomes in the in history, students from the last three to five years, whether they engage with the university, whether they can't they apply to the university, whether they were admitted and whether they ultimately came. So we have the outcome of student came, student did not come to the University. And now we have all the attributes about those students that you mentioned, you know, their SAT scores, their location, some of their behavioral patterns, did they come on the website or not? That was all pulled together and trained in a model. And the model then sees all the permutations of all that data, and is able to say, Okay, now that I know what happened in the past, I've learned and so now you can give me a future looking scenario, have a student that hasn't made a decision yet, I'm going to use all the data, the many variables and, and situations about that student based on what I've learned in the past, and I'm going to come out and tell you the likelihood that that students going to attend or not.
Greg Dickinson: Wow, wow, that's cool. That's cool. So you're so you're using? Yeah, I like that, that the analogy right of the of the past, right, you can say, Hey, we know, last three to five years, you know, 20,000 students applied 5000 attended, what was the difference between the 15? And the thought, right, and looking at, you know, and I'm guessing john, the more attributes to more data elements, you have that the better, right, because you can train that, that that model and and and begin to help, you know, predict.
John Abbatico: Exactly, right. Yeah, models get the models get smarter for a number of ways, you know, that the technology gets better. But more more than that, than any other way the model gets smarter is that additional data and behavioral data is added to the model. You know, there's, there's some scientific principle that if you if you knew all the particles in direction, in the universe of, you'd be able to predict everything that happened from now until the end of time? Well, it's kind of the same thing with data when you think about it, a 17 year old kid, if we knew all of the data about that, that that particular individual, their behavior, you know, their their study habits, etc, we would get better models to be able to predict their outcomes.
Greg Dickinson: That's funny, though, because I'm what you got. You don't you don't see me typing, I'm smiling, right? Cuz I'm thinking about you and I right together demoing and and even though we didn't have predictive analytics in the room, we were kind of automatically in our heads going, that guy over there is not paying attention. She's really engaged. Why did they ask that question? Right? In a way we're trying to predict whether this, you know, customer, would this this prospect to become a customer and it was funny, the other day, I was on a call with one of our customers, VP of Sales said, Listen, I don't forecast transactions anymore unless they have Omedym. And what that means is they engage in our digital platform and taken and used our, our technology here at Omedym, as a prospect of suspect or customer to engage digitally in a demo or a sales presentation or a message. And he said, if they haven't engaged digitally, there's no way they're buying so when my team forecast things. So he's using you know, in a very rudimentary, like kindergarten level, the behavior of the prospect and what buyer intent they display to predict whether they're going to going to buy so it sounds like down the road. We should talk to John Abbatico about how Omedym could develop a predictive analytics engine because we'll know based on buyer insight buyer behavior buyer, you know, intent, and if they if they bought or not right, we have that same analogy. You just said Did they go to the university or not that yes and no behavior than helps you train right a model? Am I am I am I putting my putting the dots together here, John?
John Abbatico: You absolutely are Greg. And in fact, the human mind works a lot like the way our models work, we use past knowledge. The models use past events and data to come up with inferences or predictions in the future. The thing that's different about these models that makes it exciting and a bit scary is you know, in your scenario, the your sales VP use one data point, whether they engage in Omedym or not. And that's a very, very telling data point, I'm sure, but the models use 1000s and 1000s of data, right. So it it think about you know, it thinks beyond what the human brain can look at and and with what that allows models to do, Greg, and this is really important is to individualize the predictions. Because every prospect is going to behave differently. I mean, there's patterns, of course, you're gonna see patterns. But the more but when you break it down, every prospect, every human is different than the next one. And so the more you know, and the more data you're able to analyze about that individual, human or prospect, the better your likelihood of predicting the outcome of
Greg Dickinson: Wow, well, I didn't I never thought of it that way. But you're right. I mean, every, we may all have, you know, similar characteristics to buying a car but but behind the scenes, there may be other data elements of Hey, this person was really interested in safety, this person was interested in 0-60, this person was interested in warranty. And they all took a test drive, but the but but the other data attributes of that critical car led them to that demo, if you will, the product demo the test drive. So you're, you're hitting on the fact that the more you collect, these engines can just run right in and kind of like it's kind of like the automated you know, the the chess player, right? The machine would beat Greg's, but 1000 times out of 1000 times, right? Cuz my, I can't think three moves in advance, much less 12 or 20. Right? And that's what these machines can do.
John Abbatico: Right? But I mean, if you that that chess analogy, the computer loses a bit at the first first couple tries, right? Because it's still new and learning the data. And then it gets better. And that's kind of what the models do, as it gets more data and more outcomes to determine, you know, to get essentially, to get smarter.
Greg Dickinson: Yeah. So john, you know, there's a lot of hype right in the world, right? Ai, ai, ai and predictive analytics. Are there are there any kind of like healthy options? Are there any myths that you could bring to their attention, say, listening to you, you know, if you hear this, this or this kind of take it with a grain of salt, right? It's in some areas, it's high. What, what are some of the, you know, the things that you know, at you being the rocket scientist, and you're living it every day, you've developed technology that helps universities with their enrollment based on predictive analytics? What are some of the things you even read and laugh? And go? Gosh, yeah, that's not true. That I think it'd be great for us to kind of hear the other side of the coin.
John Abbatico: Yeah, you know, so when it comes to the predictive analytics and AI side, you know, there's always that joke that the computers are going to take over. And, you know, I'm not going to debate that, but who knows what the future holds, but far past our time on the planet. But I think a myth that is in the short term is that, you know, predictive analytics are going to take over my job, predictive analytics, are going to replace people, predictive analytics, are going to be a lot like reporting the fact that it's going to augment decision making, it's going to make you smarter, it's going to help you make better decisions, it's going to increase your batting average. Yeah, I can't think of a better sports analogy to predictive analytics than then baseball, you know, think about think about if you can turn a 250 hitter to a 350 hitter, that that that percentage increase in decision making for that hitter to swing it, the right pitches, is incredible value. And that's really what predictive analytics is going to do and is doing in the economy, it's making better decisions. So people grow right, a lot more of the time. But in a lot of cases, humans are still involved in that decision making process. And they probably will be for for a while. Now. There are some of our customers that are, you know, get comfortable enough with the models that allow it to make automated decisions, but most are using it in decision making.
Greg Dickinson: I love that because you know in the beginning I said, you know, give, give us some time we will tie it together. And what you just said is that, you know, your customers in the, you know, in the administrative area of college universities right there, there, it's not a zero sum game, there's now nobody in the rooms anymore, it's now this big computer with flashing lights. So you know, that's figuring out which students are going to come it's the same people, but what you've done is you've given them, you know, instead of a hammer when you're building a house, and now the nail gun right there, you've made their their life easier, their job better, they've, they're smarter, right, they're more informed, they they're able to look at, you know, the prediction and probably make the experience better for themselves and right, for the students that are coming, because now you can focus more on the ones you you know, out of 20,000, you talked about, we know these 5000 are most likely going to come let's focus on the 7000 or 8000 they're in that set and we can make that experience better for everybody.I mean, it's it's making it better for all making it more quote unquote, convenient.
John Abbatico: And that's exactly the one of the biggest value props that we have for our universities is better resource allocation and how do we make it how do we make the decisions you make result in your goals? How do you achieve your goals, and University have a lot of goals when it comes to enrollment? You know, it's not just about getting the number of kids you know, a lot of universities want to make sure they get a diverse set of kids. A kids that Have a propensity to do well and succeed through the four years. So the quality of kids, firstly, geographic diversity, all those things are made really difficult when you think about building a class, and we make it essentially more convenient for our customers to build those classes.
Greg Dickinson: It's so it's so interesting, john, so think about Omedym for a second, right in the sense that we, you know, let's I get sometimes I'm a pre sales engineer by trade, right? That's what I how I started, once a pre sales engineer always one yet, there are some will say, well, geez, Greg, your technology automates the demo, right? We use video. And we allow that video to be searchable so that people can find those quick segments and snackables of of the demo, the workflow, the analytics and predictive analytics, the the onboarding that that they want to see and some kickoffs and say, Well, you know, you're trying to replace the engineer. And I'm saying, No, no, no, the the buyer today will not give all their time to the sales team, that pre sales team. In fact, the data shows about 70% of their time is spent with the seller, the rest of time is on their own. So we provide a way to make that connection in a digital way to that buyer. And also, the pre sales people don't have to give those as I turn them show up and throw up demos, the ones that are just tire kickers that are never going to buy the software now. They're not qualified. And yet you have to go out and build resources to accommodate those demos, why don't you automate those demos? and john, you and I both know that that's the worst demo we ever had to give right? Or, or you know, or booth duty, right? Where you had to go and just spend the whole day demoing show up and throw up right?
John Abbatico: Oh 100% aggravating. Those those demos where you go in blind and you don't really know what the prospect is looking for what the value prop is, the what are the, you know, the, the most rewarding and like you said, I find it might it might be difficult to remove these scenarios is when the you know, what the prospect wants there, the value prop and, and, and, and their pain, then that's where a good demo engineer gets in and says, Okay, I know exactly their value, the value, I need to show the pain that they're experiencing, and I can build a demo flow that's going to knock it out of the park. Right. Right. And those are the ones that are rewarding and where the real value is. Those Monday morning. Yeah, and we we've worked with some pre sales engineers to say no Monday morning, Friday morning demos, and and that, that show up and throw up demos. And it's sometimes you know, if you can automate it, and and and automate those or make them self service, I mean, that that 100% is is a big win for the demo engineering.
Greg Dickinson: It doesn't replace the face to face, right. Just like you know, just like predictive analytics in the in the higher ed space is not saying hey, you're no longer need anyone in the admin staff, we'll just put this box in there. We're making the same point. But you know, the fact is that they're thirsty for ways to do their job better. And the buyer is thirsty for being able to do more on their own right I mean, think about the way you buy the way I buy we don't go to a car lot and say hey, please mister mister missus salesman, tell me everything about the car, we can always work on our own. Well, how about doing that in an Omedym portal, seeing the product experience and product, getting the value prop the messaging, understanding that and then when you're ready to engage, you raise your hand, you engage your sales, you're smarter as a buyer. And guess what? Because they have access to all that data. So is the the seller. So I think there's some, you know, again, I think we tied together. Last question. I mean, two more questions before we run, john. And I, because I want to respect your time, and that is that, um, you know, the other thing I think Omedym is meant to think about here in the short term, right is is this idea of the fact that you know, that that sales person is, you know, with their mind looking at the activity of the buyer and saying, hey, these transactions look better than these transactions in their pipeline. But you know, I think we'll be we'll be get with you someday to be able to say, hey, because we have, you know, all kinds of data, what do they watch? What questions do they ask which topics and they invite somebody else, as they do it on a Wednesday versus a Thursday? Do they do it at night? Or do it on a weekend? What's their title? What's their role? While these are just a bit a smidgen of attributes we have, we should be able to write out I think, from what I learned today, you'll fold those into a predictive analytics algorithm and be able to help sales people know I have your 100 opportunities. These are the 20 that are most likely to buy in the next month or two months, pretty much based on right on these on Omdeym lists. I mean, it sounds logical to me that we should be able to put that out, you know, based on the data we collect, is that is that logical?
John Abbatico: It is and I want to take it one step further. We didn't talk a lot about prescriptions, but not only Greg, will you be able to give them insights into what may happen with some of the prospects. What's really exciting when it's extended to prescriptions is you'll be able to recommend action. That will help close those deals. So, you may see a pattern where you say, okay, you know, based on what we know, you can increase your the likelihood of closing this deal by 20% if you provide them with an updated price sheet or with a document about how this certain functionality works, that to be able to prescribe the actions that promote a better outcome is really, really exciting. And we can do that too, with the data. Okay.
Greg Dickinson: Well, that has to be called too, because that's, that's, you know, that's the cat's meow. Right? I mean, we've got again, we've got another customer that said the other day to us, he said, he said, Listen, we we no longer email the price you we put on Omedym to him. So we know that A. They logged in and that shows that they're interested B. Do they go? What do they read the proposal at all? Because we can track right every single word that they scanned over read or watched? And they they said that's another buy sign for us that they you know, because listen, john, you and I both semies in pricing, right? And 80% of time, we're going to throw the trash because we're not interested. It's column fodder. It's not really who we're interested in. So that prescription helps, man, that's really, really fascinating. It's a cool world, man. I think I think you guys have had phenomenal success there in, in getting universities and I'm sure that was a hill to climb, right to get people to understand that you're you're helping and augmenting with this idea of of, you know, more intelligence.
John Abbatico: Yeah, it's still I we've been in the business for five, six years. And it's still an emerging market. I mean, it's still understanding what your the the machine learning does and why it does it better is still it's still an it's an evolution, not a revolution.
Greg Dickinson: Got it. Hey, John, great lesson today for that for the for the audience. And on predictive predictive analytics, but I always like to close with, if they're at the gym, they're listening to you in the car, or listening on a plane, listen to you at home. And they can they're going to leave here and go back to work. What's the one? What's one thing that you could say to somebody go do this, it will make your job better your life better? Whatever. What's that recommendation from? From John about a called a rocket scientist?
John Abbatico: So, I mean, you've been to the theme, Greg, really, I mean, I think you know, this, and I'm sure all your listeners probably get a sense for this, that predictive and prescriptive analytics are going to change the way that that we all work in that we we all behave and so get ahead of the curve, because it's still emerging. There's still, there's still a lot of companies that haven't adopted, predictive and prescriptive analytic strategies. So I really strongly recommend is start to understand it. And and the one thing you got to have good data, tend to do make good models. So you know, if you're in think about a data inventory, make sure you're collecting data in a way that you can eventually use it if you're not doing predictive analytics now, because you will need it. So and then my other advice is, have your kids get into data science, because it is an incredible one of the reason set up that Othot is doing so well is because these data scientists are so darn expensive, and so hard to find it's a really lucrative field. So I would, I would really recommend you'd steer your kids that way. You'll be set for life.
Greg Dickinson: Beautiful. All right. Hey, john. Thanks for joining us today. And we'll talk soon. Yep.
Transcribed by https://otter.ai