EP14 – Big Data & Geospatial Analytics in Media & Telecommunications
In this episode of On Point I sat down with Ben Edelman, formerly Vice President of Syndicated Telecom at Kantar, a worldwide media agency focused on providing data insights and strategy for brand development, and now a Principal Consultant at Dun & Bradstreet. Ben faces the challenge that many companies are facing today with dealing with the barrage of data and being able to geospatially analyze it faster, for both marketing and competitive intelligence, specifically in his world to understand the potential of 5G and wireless internet access. And as you can tell, he’s right in the thick of it. Hear more in this episode of On Point with Korem.
Joe Francica: Ben, thanks a whole lot for doing this. Really appreciate your time. Let me just jump in and start with an overview of Kantar and maybe your particular focus in the telecommunications industry.
Ben Edelman: Absolutely. Yeah. Thanks so much for having me, Joe. I really appreciate it. Yeah. So, I work for the Syndicated Telecommunications division of Kantar. Kantar itself is a very large multinational advertising market, and consulting firm that does business with many, many of the Fortune 500. It helps them with everything across the board. Syndicated Telecom is a little slice of that pie where we do primary research revolving around the likelihood to buy telecommunications and IT products.
That’s been our bread and butter for many, many years. And the output of that is some predictive scores that allow for companies to do things like market sizing and just other market analyses in the marketing of their products to businesses. My personal focus has been on the B2B side, but we also do work on the residential side as well. So, that’s pretty much the top line stuff Syndicated means, we sell that research across many, many customers.
JF: Across internal customers are obviously external as well.
BE: External. Yeah. So, there’s a little bit of use of the data internally to Kantar to help understand the telecommunications industry. But the big use cases are with really the tier-one telcos and understanding their footprints in their market.
JF: So, maybe let’s go into that a little bit about the kinds of applications those clients are looking for, kind of the pain points that they’re challenged with today.
BE: Certainly. So those clients, they really need to understand the network landscape, both physically, geographically, and technically as well. You know, people are using more and more data than ever before. And they’re using it in new and different ways, 5G of course, being a big one.
And so, our clients need to understand who’s going to want the data, where do they have to lay infrastructure in order to satisfy the demand. And it used to be much simpler where you simply knew that this customer was here, and they probably needed the Internet. So, you build them a DOCSIS line, and they’d be happy with a baseline SLA. Those tastes are evolving among customers, especially with the cloud and outsourcing of technological products.
JF: So, let’s go into that a little bit about 5G. I mean, to build out the infrastructure, there’s a lot of talk. Are people adopting it? What is it that I think that Kantar can help clients with specifically on 5G?
BE: So, we have access to huge swaths of data that help us understand things like foot traffic, that help us understand who’s demanding what, from a bandwidth perspective and both a business and a consumer-level so that we can actually geographically show this area has X demand for data, both wireline and wireless.
But it has this underserved, for example, bandwidth needs. So, that’s a flag for a telco or an MSO to be able to go in and say: ”All right, there is demand, we can actually meet that demand” and thus, but our products and our network in here to do that.
JF: What’s the, what do you think the impact will be from the infrastructure bill that was just passed in the U.S.? Do you see an uptick in a need for additional analytics and services to the telcos?
BE: I think a little too early to tell on the infrastructure bill. Historically, when there’s been investment in say rural broadband, the end results of those government investments are three to five years out from the passage of a bill. And it’s a lot of time and effort is spent sort of figuring out how to get ahold of that when he rather than the analytics surrounding where the builder needs to happen, that tends to happen after the cash has been secured.
So, I think it’ll be a boon for the industry. I think it’s important that infrastructure investment is made with telcos and MSOs. I mean, the last two years have demonstrated that the Internet is just absolutely critical for day-to-day life at this point. And so, to see some government support for that, it is heartening. Yeah.
JF: Yeah. So, let’s get a little, it’s a little bit about the geospatial analytics that you’re able to do. Maybe, just give me an idea of the kinds of applications that you’re engaged in. Like I said before, I’m always interested to hear what new things companies are doing with geospatial and maybe where you want to take it into the future.
BE: Absolutely. So, we have what I like to term it, something of a game-changing product. The holy grail of the telecommunications industry for the longest time has been the ability to understand not only where your network is, but where your competitors’ network is.
And being able to tease that out from datasets has been exceedingly difficult. I like to say until now. You know, I’ve been working with Korem specifically on this for the better part of a year now, and we have something that’s pretty much ready to go where using things like building polygons I can actually take geospatial data points and sort of do a spatial join where I can see, you know, this internet provider is at this point. Well, that point falls in this polygon. Well, now I know as of this date and time, this entity is serving this physical structure
That’s never been possible before. You know, best case scenario. Most providers say, well, I see a Wi-Fi signal here, and here’s the SS ID. So, I’m pretty sure they have internet. Now I can tell you who’s there. And it just, it would not be possible without things like the building polygons that Korem is providing.
JF: So, what does that mean in terms of as a media company, what you’re able to serve now that you have all of the data available to you?
BE: Well, we’re fairly siloed here in Syndicated Telecom. So, I am reaching out to my colleagues in media and retail and saying: ”Hey, we have this pretty cool thing.” Those interviews and chats are happening as we speak. But from a telecommunications perspective, you know, our clients are interested in exactly this kind of thing. And so, it’s really been an end customer as opposed to an internal customer-driven product to this point.
JF: So, is the analytics you do, is it a lot of spatial join spatial clearing, and maybe what’s the output that you would show to a club?
BE: Oh, for sure. So, the spatial joins are happening very much internally. The end client would probably get something of a flat-file with perhaps a building centroid. You know, I like for them to be able to see the polygons, but they’re not all quite advanced enough to make use of that just yet.
But so they’ll get, say something of a flat file, it’s just for like, here’s the location, here’s the ISP at that location. Here’s a secondary or tertiary one if I’m able to prove it. And here’s the daytime where I was able to prove that from. And so, they can take that and they can say: ”All right, here’s to my network and oh, look! Here are nearby my competitors’ networks. And here’s how that’s evolved over X period of time.” So now they can know: ”This company is doing a build-out in this city. This company is doing about that in that city, I need to be on defensive or offensive to maintain my share.”
JF: Yeah. And how do those services break? I mean, the services themselves that you might suggest because you know the location. What is it that’s revealing to the end customer?
BE: So, it’s competitive intel, right? And many of them, I would say most, if not all major ISP at this point, are focused on fiber and they want to understand where the best place to build their fiber assets are. Are these fibers expensive to build and very fragile? So, it’s difficult and it’s a very different way of building out from the previous infrastructure that involved, it was called DOCSIS is the cable modems you might be familiar with. And so, understanding if there is a fiber provider in a building already, that’s going to save you a ton of time.
Building a second fiber line into a building is like building a second electricity building. There’s just no point. So, if you know where the other fiber buildings are, then you can say: ”All right, let’s cross these off my list. And oh, look, this building doesn’t have fiber. Let’s go talk to the building owner. Let’s go talk to all the potential clients in the building and see if it’s worth building that out.”
And that’s really where sort of our bread-and-butter predictive scores come into play. Because now we can say: ”Here are all the businesses in that building here are the bandwidth needs that they have.” You know, a 50-person manufacturing company is going to have very different needs than a five-person video editing company.
JF: Right. Right. So let me ask you about the predictive scores that you’re developing. I mean, without maybe revealing some proprietary information. How do you do it? I mean, what kind of either spatial analysis or spatial statistics are you employing?
BE: So, for what it’s worth, this is just good old-fashioned elbow grease. We got our teeth in the research industry, so we pick up the phone and we call thousands upon thousands of businesses every year and we have hour-long interviews with the business decision-maker and say: ”What do you spend on this? What do you spend on that?”
But the cool thing is we actually have the specific de-anonymized data for each one of those interviews, so, I know physically where that business is located. And with that, I can actually do geospatial analysis to say: ”All right, what are they paying in Texas? What are they paying in New York? What are they paying in California?” And that kind of understanding is really helpful for larger ISPs because 750 feet of infrastructure is very different in rural Texas than it is in the thick of New York City.
JF: Yeah. So, that’s really interesting. So, let me ask you a little bit about that. How much more data you need at this point? I mean, we seem, you know, location-based data is exploding. I mean, if you had your meadow goggles on looking into the future, what more do you want and need? We work with a lot of different data providers, but you know, there’s more coming. So, I’m just curious. What’s in your crystal ball at this.
BE: It’s a completeness question, right? No dataset is perfect. No dataset is completely full and getting those edge cases handled, those are going to be the places where the last little bit of money is to be made in this industry.
If there’s a polygon that’s missing or a rural town that has really poor GIS, you know, I rely on my special vendors to go and see if they can tease that out through whatever mechanisms they have available, because I can’t go and hash that out myself. But as the data sets get broader, get deeper and more complete, my end analysis becomes that much sharper.
JF: Right. Right. Let me ask you personally, how you got interested in space? A lot of people, they come into school, they get GIS degrees, but sometimes it’s self-taught right? So, I’m kind of curious how you got into this.
BE: So, that’s a great question. Thank you. Yeah, I do not have a technical credential in spatial analytics whatsoever. I’ve been a user of a software called Alteryx for many, many years, and within that software, there are spatial operations. For the longest time, I did not understand them but I did go through a couple of the trainings that they have at their conferences, so, at least, I had a baseline grasp of it.
And it was early this year, when I was introduced to an absolutely massive data set surrounding those ISPs that I sort of understood, ”Well, if I have this dot and I can put the dot in the polygon, then I can actually empirically tell what’s happening at this location. Well, I really need to understand how to do this.” And so that became the impetus for drinking from the spatial fire hose.
JF: Yeah. Yeah. So, have you found Alteryx a fairly intuitive tool to use and does it do what you want it to do, you know, in terms of linking, or it could be massive data sets together and then getting to an answer?
BE: It gets me 80 to 90% of the way there. It gets me to the proof of concept. And that’s really cool. And so, if I’m having… my personal record for an Alteryx workflow was earlier this year with this particular project and it took 157 hours to run, and that was purely to analyze Cook County, Illinois, in the city of Chicago, and the surrounding suburbs.
And that kind of leads to, you know, ”What do I do next, right? If I’m having the data at a velocity, such that all of our systems running constantly 24 hours a day, can’t keep up with a single day of data, let alone months or years of it. And what do you do?” And so, coordinating with, of course, my friends at Korem and a couple of other geospatial experts, we ended up offloading much of that to a cloud service. And in that, was able to handle just monstrous amounts of data.
JF: I don’t think you’re unlike many users who are now running up against this challenge of larger and larger datasets. You’re all flirting some to the cloud. You’re doing some, you know, in a solution such as Alteryx. It seems like you’re going to continue in that vein only, and the need for more and more data is just going to get, I would think more, I guess, at this point.
BE: Absolutely. And that, not only is the data just going to be coming faster and faster and more and more. I need to be able to crunch it, not necessarily in real-time, but very, very quickly. And these are all pain points that I’ve never had to deal with before. Previously, I could run a couple million records in five minutes and call it a day. But at this point, I’m looking at 30 billion records a month, then I’m having to spatially join to 142 million polygons, you know, do the math on that. Unless you have efficient algorithms, you’re gonna wait until the heat death of the universe for it to complete.
JF: So, I would assume the next step is maybe using some of these more cloud-native solutions that are handling this. Again, without revealing much, are you looking at those today?
BE: Absolutely. Yeah. I have some trusted vendors that are taking me through the various options. You know, the bigger players tend to be the better suited for it, they also tend to be more expensive. So, it becomes a trade-off. I can say we evaluated one particularly well-known major provider earlier in the year. And our consulting team was able to cut the price by a factor of 90% between what we were being asked for. And so, that meant the difference between doing the project or not.
JF: Sure. Sure. So maybe just one final question. You know, technology’s evolving rapidly, you’ve got a challenge with big data. How can we help you more and more, I guess, this is the bigger question of… and we’ve obviously provided some data and consulting. So, I guess that the next step is just relying more and more on some of these cloud tools.
BE: Absolutely. So, the cloud tools are great. And one of the things I’m looking forward to is sort of the vertical aspect, like the literally physically vertical geospatial aspects of ”I know some of my data is going to start coming with altitude. How can I put people on the floors of buildings as opposed to just inside of a polygon?”
And for what it’s worth, that’s not something Alteryx is necessarily capable of. Right. So, I’m going to need to lean on some of my geospatial experts to help me in 3D space.
JF: Yeah. Yeah. Well, and that’s always the interesting question for us is, we live in the 2D world a lot of times, but 3D is now becoming such an important aspect for sure. Yeah. Well, Ben, thank you so much for your time, really appreciate. Obviously, we value Kantar as a client. So hopefully we’ll continue our relationship.
BE: I look forward to it very much and an honor, and a pleasure. So, thank you so much for having me.
JF: Thanks Ben. Thanks again for joining us on another On Point with Korem. And if you like today’s podcast, please leave a comment in the comment box where this podcast is posted, which could be Apple Podcasts, Google Podcasts, Spotify, or YouTube. I hope you’ll join us next time for another On Point with Korem where we’ll get On Point.