...the only data worth looking at really is the data that's hard to get...that's really why we need to look at it...
Michael talks us through his philosophy on making the best use of data - that data is a lifestyle, not a project.
Ben Newton: Welcome to the Masters of Data podcast, the podcast where we bring the human to data and I'm your host, Ben Newton. Our guest on this episode comes from an industry that has long been a pioneer in data and making good use of data. Michael Herskovitz is a SVP and co-head of global operations and technology at AllianceBernstein. AllianceBernstein is a global asset management firm providing investment management and research services worldwide to institutional high net worth and retail investors.
Ben Newton: Michael is going to talk us through his philosophy on making the best use of data, which he's built over years and years of experience in the area, that data is a lifestyle, not a project. So without any further ado, let's dig in.
Ben Newton: Welcome everybody to another episode of the Masters of Data podcast. And as always, I'm excited about my next guest. He comes from a realm that we haven't had a chance to talk to somebody about in banking and investments. So I'm very excited to have Michael Herskovitz here and welcome come onto the show.
Michael H.: Thanks, Ben. I'm really glad to be here. Thanks for inviting me.
Ben Newton: Absolutely, absolutely. And Michael's over at AllianceBernstein, he's the SVP Co-Head of Global Operations and Technology. So definitely a lot of application of what we usually talk about. And as always, Michael, I love to hear people's background and you particular, just a little that I could see kind of coming out in some of the sources I found, I think you definitely seem to have a pretty interesting background in terms of how you arrived at where you're at. So what got you, you know, number one into banking investment and what kind of led you towards this ... It seems like you kind of have an analytics background too. So what led you that direction?
Michael H.: Well, you know, I got started in finance many, many years ago. As a teenager, I used to work in a pawn shop in Baltimore.
Ben Newton: Really?
Michael H.: And it was very old school. I mean pen and paper, filling out forms by paper and learning about everything from outboard motors to diamond rings. It was quite a fascination. But along that lines, you know, when I was in high school, we had a computer science, computer math course. And in the entire county is Baltimore County, outside Baltimore we had one IBM 1130 computer full, you know, blistering with 8K of memory. And [crosstalk 00:02:27] I had was that our computer science teacher was really into gambling.
Michael H.: So he had-
Ben Newton: Really?
Michael H.: ... all of these gambling theories, you know about, you go into a casino with $1,000 and you start bidding $2 in roulette on black. How long does it take to wipe yourself out or changing things to like Fibonacci sequences.
Michael H.: And it actually made it interesting and we used to write all these programs to do that. But at the same time, you know, the PCs were first coming out, the original Radio Shack TRS Trash-80s, and I had an aunt who was a professional astrologer and she had me helping her with her charting formulas for astrology, you know, saving things onto cassette drives. Those were the fast media at the time. So I kind of got into this, you know, programming sort of the hard way just by learning and doing, but then went on to school, went onto Carnegie Mellon.
Michael H.: And at the time they actually didn't have an undergraduate computer science program, so you had the applied math. I did this mix of math, finance, computer science, helped pay for my education by working as a programmer and went straight through three years of undergraduate, and then on to their a graduate business school. Really then after a couple of jobs bounced around, found my way to Wall Street and started working in this aspect of trading analytics, hedging, which was just getting off, just sort of taking off at that point and working on a corporate bond trading desk, Merrill Lynch, which at the time you could still smoke on the trading desk.
Michael H.: That's how long ago it was. And we had the cathode ray screens that threw off enormous amounts of heat. But anyway, they would try to hedge their positions. They wanted essentially to minimize their market risks. They had the run all these regressions of price variations against your hitching instruments and hope that it worked. You know, because if you had these traders who just really would come after you. Then, they lost a lot of money in mortgage backed securities and asked me to go over there and help work on the research side, which was just perfect for me.
Michael H.: It was my first experience really in this mixture of large amounts of data, math, computer science, because you use this data about people's prepayment behavior to try and understand as interest rates moved, how much were people likely to prepay their mortgages, how that impacted the value of the securities.
Michael H.: You had to write all the things that required bond math, but all these econometric formulas to try and anticipate people's prepayment behaviors. And this was probably the year where it was kind of big data, but we didn't know big data at the time. It was fascinating because you could actually write models and value securities, create new things. I spent a long time looking at that stuff.
Ben Newton: When was that? So this is like early 2000s, or ...?
Michael H.: This is late 1980s, early 1990s.
Ben Newton: Okay.
Michael H.: It was, you know, an early year at the time. When I first got there, we were still doing all this work in mainframe computers and making the transition from interpreted languages. APL was the big language they use then to compiled languages and then on the client server world, so living through it all. But this aspect of understanding how to manage the data and our data used to come in that monthly prepayment information actually on physical tapes. It was basically a race for all of the firms, it was Merrill Lynch, Solomon Brothers, Goldman Sachs. Who could get their tapes loaded, data aggregated faster and crank out what actually happened last month?
Michael H.: Because that gave you a competitive edge to just what happened as well as to understand like what behaviors were happening at different types of pools and why certain people were prepaying faster than others. It was really cool.
Ben Newton: That's intricate. I mean one reason why I was asking you about the timing is you talk about big data, but I mean you're talking about big data. Well before anyone would have labeled it that. This is kind of like the dawn of big data. That's cool.
Michael H.: We used to do these massive simulations and we would, at the time, look to rent time on super computing centers, controlled data, and cray, and I remember visiting the Minnesota super computing center at the time, and I was more impressed by the fact that in the winter time they used the heat from the cray to heat the entire building. So, it gave them a huge incentive to have massive uptime for the machines.
Ben Newton: You mentioned the cathode ray tubes being so hot. Is that where the term boiler room came from is because it was so hot sitting in front of all those screens?
Michael H.: Nah, the boiler room was more where you've got people that are promoting stocks and trying to get [crosstalk 00:07:10] .
Ben Newton: Oh, okay.
Michael H.: Hey, man, you better get in on this right now. This firm, you know, they have this great hair transplant formula. It's going to go public.
Michael H.: How many do you want to buy now? No, the boiler, this is more the days of liar's poker types of things from the original Michael Lewis book, which I think would be a great movie.
Ben Newton: Oh yeah. Absolutely. So you were getting into the data early. I mean you're kind of on the applying analytics to those data sources and getting into AllianceBernstein where you're currently at and basically it was in the early 2000s is when you headed over there?
Michael H.: Yeah, I joined in 2006 right before the financial crisis. I went through a couple of ... went through this transition from being a researcher, mortgage researcher, to just working in technology full time. You know, I looked at all those mortgage problems or challenges for many years, really did everything I wanted to do there. And then the whole revolution in object oriented programming, client server technology was taking over. And I found that that was an area where, because I started out working on a trading desk, understanding the business that I could really be a good manager and understand how to get the technology done and developed.
Michael H.: So I worked a little bit on what we call the sell side than the buy side, moving that to AllianceBernstein in 2006. That was really interesting because you know, one thing that's interesting about AllianceBernstein and one of the reasons I enjoyed working there, is it's got a very heavy research culture and I think that that has been something that requires like trying to look at all sorts of data, a fundamental economic information of quantitative information and looking to try and understand how that can be used to sort of form views of portfolios.
Michael H.: And you know, I think that this is an area that certainly is well ... I mean there's lots of people, a lot of smart people are looking at these, at data over time trying to get ahead of what's happening in markets, looking for patterns. It's an interesting intellectual challenge but also it's interesting because for your clients, the people that are investing, which could be anything from, you know, you've got your 401K plan, to an insurance company, to a mutual fund holder. You know, you're trying to work to the benefit of the ultimate investor, which kind of gives you a sense of like, I'm doing this for a purpose, you know? I'm not just doing it an edge. I'm not doing it to rip people off. I'm trying to do it because I want to make money for my investors.
Ben Newton: Yeah, it's fascinating. I remember when I was doing my own research. I mean that kind of commitment to research and analysis comes across pretty clearly in AllianceBernstein's, you know, website and everything they say about themselves. I think that's pretty cool. And in a lot of ways I would guess that, you know, your industry has always been on the forefront of data to a large extent. Because you have to be.
Michael H.: It's the lifeblood, and it's going through this transformation now where for a long time it was, you know, fundamental economic data, company data, but now it's making this pivot towards using better analytical data to understand who your clients are, who you could be selling to, what markets make more sense than others, as well as the aspects of alternative sources of data. And I've often often told people that, you know, we spend a lot of time trying to forecast, say, what auto sales will be next month or next quarter, next week [inaudible 00:10:26].
Ben Newton: Right.
Michael H.: But you can actually be much better off if you actually know what's happening right now. Like if we knew what auto sales work today, you would actually have a much better idea of what economic conditions were than understanding what they were two weeks ago by trying to [crosstalk 00:10:44]. And I think this thing where you're looking at alternative data sources to try and understand what's happening today is one of these things that generally gets overlooked.
Ben Newton: Yeah, that's really ... You know, you actually reminded me of something that I've had conversations with people about and something that I think has been interesting is this idea that it's a combination of data and analytics that really brings this to life because we're awash in data, but it's getting the right data and it's applying the right analytics to it that's really what's going to make you competitive? It's like, you know, one term I've heard uses idea of an analytics economy, but the idea is to companies, the firms that are able to not only get the data and get real time data, but also apply analytics to it. Those are the ones that are going to win. Those are the ones that get ahead. It has to be a combination of all these things.
Michael H.: Yeah, they're the ones that are winning. Now, one of the analogies that I use, you know, comes from actually advertising. There was a department store had John Wanamaker who had this thing where he said, "Look, you know, spend like 50% more on advertising than I know and I should, but I don't know which is the wrong 50%."
Ben Newton: I'll remember that.
Michael H.: [crosstalk 00:11:49] data. Like, we have more data than we need. You know, when we [crosstalk 00:11:53] multiple sources. I tell people we got 150% of the data we need. We just don't know which 50% to get rid of, and it's [crosstalk 00:12:01] dated and you know, messed around with. That usually turns into ... and then there's always this push to like, oh we need to buy a little bit more and we need to buy a little bit more. And that just compounds your problems. If there were a data hoarders show, that would probably be interesting too.
Ben Newton: Yeah, yeah, yeah. Maybe that's the next idea. We'll work together on that. Yeah, no, absolutely. And you know, and it's one of those things that there's always a sense of if you end up going down some sort of investigative path or you're trying to understand something and then you end up not having the data because you didn't get it, that's not the conversation you want to have. Well, you know, I think you've talked to a lot of like how you guys leverage data and what you're doing and it's always super fascinating.
Ben Newton: But you know, when you in particular, and we'll put this article when we post the episode, you know, a great article you wrote about this and you talked about, you know, good data is a lifestyle, not a project, which I thought it was a great idea. And you know, what's interesting to me is like, so with all the background over that you just painted, why did you feel the need to write that? Like what was kind of driving you? What kind of misconceptions or problems were you trying to confront that made you kind of develop those ideas?
Michael H.: I probably had gone through another one of those meetings where you're sitting around and one of your colleagues saying, "I can't do anything. All of our data's terrible. I don't know what to do. It's horrible. We've got a huge data problem." And it was like, I've heard this. It's like deja vu or it's just another groundhog experience. And when you sort of press people like, "Well what is it? Which data do you actually mean?" Because data is such a broad topic and it's hard to pin down, but they say, "Well, I don't know what we should do. Let's just hire a chief data officer and he or she will just figure it out. Let's just dump it on them," and somehow a miracle will happen.
Michael H.: Or let's go out and hire this company. Or I heard about this, you know, if we just put it into this massive data ... it was like let's create a data mart.
Michael H.: Then, for awhile I was like, let's create a data warehouse. And then it was, somebody came to me with the idea of like, oh, we need a data lake now. And I was just saying the only solutions that are masking and it's like no, no.
Ben Newton: So, you're actually solving the problem.
Michael H.: Yeah. And it was just to kind of think about the notion of what has worked, you know? In areas that I've experienced that have worked well, and it was this notion of continued focus on understanding how to get improvements and a better overall, I say, culture around like a data centric culture because you're all carrying your data from one application or one domain to another, and how do you come up with aspects, I'll call, like of golden sources or master sources of key data and ways of ensuring that you have the integrity of that data flowing through the organization? This becomes much more important when you want apply analytical constructs to the data.
Ben Newton: Yeah. How do you really go about doing that at a high level? Because I mean I think what you're getting at is key here because in some sense your conclusions in your analytics based on the data is only as good as the source data and actually applying some sort of ... actually feeling confident in your source data and having those golden sources is pretty key, but I mean what does it actually mean? Is it in terms of organizing it and processing in a certain way or just having like a shared understanding of it, or ...?
Michael H.: Well, you need to get buy in from your colleagues and whether it's on the technology data or research side. I found that the best way of doing that is by coming in and giving them examples of like crazy things that happened with your data. You're trying to understand your sales data in certain areas, but you don't really define your products in a consistent way. You know, you have to come up with ideas of how to join data across different aspects and when you don't have common product identifiers or a sales person's name is different in one system versus another, so you can't actually figure out what people are selling.
Michael H.: You put these ideas out there and you show people some of the craziness that they have. They start to understand, oh yeah, this is a problem. This is why I can't get a sales report, or this is why I can't determine like how much holdings I have of a specific company.
Michael H.: And I think that this is important and especially where you've got complicated ownership structures of companies and especially in the dead area. Like for example, you know, you're trying to figure out what is my exposure to, say, China. Now, this can be really complicated because the Chinese government for example, has ownership stakes in lots of different enterprises. Not necessarily 100%, but you don't really understand your ownership stakes and things until you actually start cleaning up your data.
Michael H.: For example, we had somebody who thought Gazprom, you know, the Russian energy company was a Luxembourg company because that's where they had issued their debt and it was like, "No. The [crosstalk 00:16:58] of this company is actually Russia. And different investment companies when they would look at emerging market exposures, suddenly they had huge exposure to like Luxembourg.
Michael H.: It's like, come on. Before you know it, you're out of Luxembourg. And so it was more of like going back and understanding what reference data really needed to be and you couldn't just use it directly, like I'll call out of the box. You needed to apply business rules to it. So you looked at these like examples and said, "This is why we need to spend time on it." And you sort of let people have the expectation that you're going to start in one sector and you're going to make progress and they start to see the improvements.
Michael H.: The problem with once they start to see the improvements is that the problems go away and they forget like how bad they were, you know? They realize it's like, oh, I'm not complaining anymore. And sometimes that's hard to prove that it's, oh, this was a real benefit because when the thorn comes out of your foot, you just sort of forget about that you were really in a lot of pain.
Ben Newton: Yeah, yeah, no, it makes a lot of sense. I guess part of this is keeping that memory and keeping that in the view of people that they understand the progress you've made. And one thing you mentioned in the article, which I thought was really great, you talked about this idea of data stewards, so based on what you said, I am guessing like data stewards are kind of the main characters here in this whole idea of golden sources, right? Are they the ones that are kind of committed to making sure that that data is what it should be? Is that the right way to understand it or ...?
Michael H.: Yeah, the data steward typically has the most vested interest in the quality of the data, the quality of the information. Now, they aren't going to understand how to load data, how to join tables, but they will be the ones that really ... so say it's a research analyst. The research analyst is going to really care that you've properly captured time series data. If it's fundamental information about corporations that it's all stated, you know, in the right currency with the right frequency of reporting.
Michael H.: And so that is an important thing. So you need to get their interest and understanding that they can't just use, I'll call, like the valet method of either technology or data where they just drive up, drop off their problem and want to come back and pick up this a week later on. They've got to set up these governance groups and have a mix of their stewardship, their ownership of it.
Ben Newton: So does this mean that you have ... I'm trying to think of the best way to state, but in some sense, I mean, you've got a lot of different data sources and the idea that you have different data stewards that kind of have some sense of ownership over the different data sources. Is that how it would work?
Michael H.: That's all it will work. So you know, you may have different stewards, some who deal with account data, like information about our clients or our products. You may have others who are really focused on quantitative information or others that are really understand how our sales data works. And those are the people ... they have the greatest domain knowledge. So they'll understand what they actually need the data for and where problems are likely to come up.
Michael H.: Just found that you can identify those people and they usually come in and they usually want to solve the problems too, you know, but they usually don't have the resources to do it or they don't really understand how to do it technically. So often what they do is they create a local copy of the data and [crosstalk 00:20:07] it or modify it there, which you know, just deals with the symptoms, not the underlying root problems.
Ben Newton: No, that makes a lot of sense. You know, because you talked about this other category of data operators and you know, what it reminded me of is I've actually had the privilege of interviewing a whole bunch of people in kind of the data science realm. And one thing, a thread I've seen kind of going through some of these is that kind of in this data science specific realm is that it's moving from just researchers playing with data and coming up with conclusions to actually having to productize the process.
Ben Newton: Like, they actually have to take what ... you know they were running on their laptop essentially and actually turning this into a production. And it seems like when you talk about both data storage, but also particularly this idea of data operators, you're talking about, you know, taking this data pipeline that is so critical to everything you're talking about and actually applying essentially like, you know, a systems management type of approach where you're actually treating it like a system and managing. Am I understanding it right?
Michael H.: Perfect. Yeah. You got it nailed properly. There's two parts of it. I'll call like the intake. So when you're actually, you know, you're pulling the data into your organization that you're doing it and you're looking for problems. Like, is the data all empty? Was it the same thing that I got sent yesterday or did they divide everything by 100 or multiply it by a million, you know? You'll see different things like that. So one is the cleanup on the intake. The other is I'll call like the exhaust part of it. Are people able to access the data quickly? Do you have like data dictionaries? Do you have tools that people can use, whether it's raw SQL or you've published tables or you have a messaging layer.
Michael H.: So, the operator really is critical in that they handle the day to day pieces of keeping that plumbing working properly. And that's an important part. Once you've got an established practice, you've got to work on that. Then, you always have to make it better, so you'll collect more, or you find different usage patterns. That's where you'll need to say, "I need to pre-stage the data, I need to cash it, I need to put it into a mark." That's where you start to feel like now we're really moving well.
Ben Newton: Yeah, no, that makes a lot of sense. Yeah. One term that I've come across recently is somebody you know, kind of similar ... There's terms floating around like DevOps. They've actually called it data ops.
Michael H.: Data ops, very similar thing, you know, which I've also heard about this term called like the data wranglers.
Ben Newton: Yeah, yeah.
Michael H.: I think those are the folks. They're kind of when you're still in the, you know, early stage of understanding like how do I clean the pieces up? How do I make sense of multiple conflicting types of information? So they're kind of your early stage, you know, triage, kind of like the first responders to your problems.
Ben Newton: Yeah. I don't know. Whenever I hear data wranglers I have some image of a cowboy with a lasso.
Michael H.: Yeah, raw hide.
Ben Newton: Yeah, exactly. Well, you know, and one thing you brought up, which is you're getting these different roles laid out and really thinking about the end to end process. But you know, it's not going to work if you don't have buy in from, you know, the management and the decision makers of these organizations. So how do you think about it? Like how do you get buy in when maybe the people that you're talking to are not necessarily going to understand the technology. How do you actually get their buy in to actually [inaudible 00:23:17] this kind of investment going forward?
Michael H.: Yeah, that's a really good question because you've got to take this thing almost like different phases of venture capital funding. You need to get enough to sort of get an initial project up and off the ground and often by substantiating like, okay, if we get this right we'll be able to maybe sell 10% more products, or we'll be able to reduce the time or potentially avoid errors. But I think the main stage is that once you are able to say. "I'm getting improvements in either quality or time to market of a certain type of instrument where sales are improving," then you say, "Okay, now I need to go to the next phase, which may mean I've got to invest more to really expand this from one product line to multiple product lines."
Michael H.: So you've got to like have some metrics, I think, that are important or some use cases who are also, I'll call, like testimonials, like you know, users who are in there to say like, "It's changed my life. I feel so much better." You know? So, part of it is like, you know, raw analytical data to support your data program and some of it is sort of the emotional part of it of you know, a lot of it could be like researchers, I'll talk about the fact that it's like they're chefs, but they spend so much time doing the prep work of getting their data into a point where they can actually do the research that they're spending 80% on the prep and 20% on the research, and if you're able to change those ratios, so it's 10% on the prep and 90% on the actual research, they're suddenly saying, "Look, I'm spending a lot more time on actually building more models or looking at more cases." And that's a huge win.
Ben Newton: Yeah, no, I mean at the end of the day, that's kind of get back to where we started is that that's what's going to give you a competitive advantage is if you can actually have your people spending more time on innovating and doing something that's unique and competitively differentiating for your company, then that's good for everybody. And I'm assuming based on your background and what you're doing, is this something that kind of based on your own experience of getting this running at AllianceBernstein and kind of, you know, been in the trenches yourself trying to get these projects going forward or ...?
Michael H.: Yeah, I mean this has been a lifelong problem I've had. Maybe it's just sort of the way I grew up. I remember in graduate school I had a professor ... it was in a, call it, cost analysis and estimation class and he once commented that the only data worth looking at really is the data that's hard to get, and that really stuck with me because when people would say, you know, "This is really hard and this is really challenging," I would always remind them and remind myself like, that's really why we need to look at it. Data that's easy to look at everybody can see, so it's usually the interpretation.
Michael H.: So I felt like using that as an aspect of like not getting discouraged by some of the challenges of when you have missing data or it's inconsistent or incomplete of understanding, like that's where you get an edge with doing it.
Michael H.: So, it's been sort of an occupational hazard of mine, so to speak, that I find myself in these situations. And what's great is that as the technology's evolved, as the tools have evolved, as now we have cloud-based resources, we have all the open source tools for data analysis, that it's like taking this to a much more interesting level of capabilities and sort of computing resources that you can deploy, which I think has opened up possibilities that did not exist 20 or 30 years ago. That, to me, is what's really exciting about the current situations we have.
Ben Newton: Yeah. Yeah. Well, I guess to kind of put a bow on this, I'd be interested to know what are the big trends that you're focused on right now that you think are going to deliver a lot of value in this area?
Michael H.: I think it's really where you take a lot of your ... two parts for me. One is understanding, you know in the world, say, of operations, how you can use data and the mix of data, machine learning, natural language processing to be much more efficient at your ability to, I'll call, it investment operations. So you know, this has followed a trajectory where many firms said okay, I'm going to move from high cost labor sources to low cost labor sources and use that just to find a way to manage the cost and margins to saying, "Look, now I can actually use much greater analytical tools." So if you've got, we'll call, like reconciliation breaks, this is where you're comparing my account, that I think I own this amount to what the bank says and try and understand why they are different.
Michael H.: Many of the reasons that they will be different will fall into sort of consistent categories. And you want to say, "How can I harvest that to solve or resolve these breaks much more faster?" Instead of saying, "Well, I can now go from this low cost area to an even cheaper place." It's like, no, that's sort of not where you want to go. You want to automate these jobs, you want to automate these tasks to a point where you don't have to do them anymore. So you have to mix of of these tools. You know, we'll see how the distributed ledger plays out. There's a lot of work that's going on now with it.
Michael H.: I think when we look back and I don't know, three, five, 10 years, it's going to be very, very interesting to see the people that will succeed have kind of done this work and have made it work for their companies.
Ben Newton: So I know ... I think that's definitely fascinating. I think you're ... one of the things I have been thinking about a lot is I think this idea of using, like you said, AI and machine learning and all these different related subjects as a way to basically amplify the capabilities of the people using the systems. And you know, I always like to think of it like an Iron Man suit. It basically makes what you do better.
Michael H.: Yes, exactly that, that's what it's about. When I tell people ... because I guess now that I'm old enough, I'm like the dad figure, people that are in school, I tell them, "Look, you know, if you're not taking a statistics course somewhere in your curriculum, you're really going to be at a disadvantage. It doesn't matter if you're a liberal arts student or if you're on the engineering side. You need to understand how to analyze and manage data." Probably the only textbooks that I go back and refer to, you know, 30, 40 years out of school or my econometrics books, my quantitative methods on factor analysis and some of the operations research. So really that persists and it endures. So my advice if you're in school, stay in school but also [inaudible 00:29:40] course.
Ben Newton: No, I think you're absolutely right. I mean, it's those kind of core concepts that really kind of carry through. But, you know, Michael, this has been fascinating. I mean, I love seeing how somebody like you that's been in the industry for as long as you have and banking and investment in that whole area. You guys are on the forefront of data. You always have been. So I thought it was fascinating to see how you think about it. I appreciate you taking the time to come on the show with us. This was great. It was fun.
Michael H.: Thanks, Ben. I really appreciate it and you know, enjoy all the series that you have. Thanks.
Ben Newton: Thanks. Everybody, thanks for listening and, check us out on iTunes, Spotify, whatever the place that you find your podcast and rate us so that other people can find us and look for your next episode in your feed. Thanks everybody for listening.
Speaker 3: Masters of Data is brought to you by Sumo Logic. Sumo Logic is a cloud native, machine data analytics platform, delivering real time continuous intelligence as a service to build grounded, secure modern applications. Sumo Logic empowers the people who power modern business. For more information, go to sumologic.com. For more on Masters of Data, go to MastersofData.com and subscribe, and spread the word by rating us on iTunes or your favorite podcast app.