Transcript

Dave: Hey everyone, this is Dave with Flyover Labs and thanks for listening to another podcast. Today, we are very lucky to have Dave DuVarney and he is the Founder & President of Talavant, and Talavant is a business intelligence (BI) firm located right outside Madison, Wisconsin. So excited to have Dave with us today. Dave, thanks for joining.

Dave DuVarney: Yeah, great, quite excited.

Dave: Yeah, we are going to do this one in person, so it’s kind of exciting, like I can actually see the patient’s face. So, I brought Dave on because this will hopefully be a good education in BI. I think, I will learn a lot, so we’re going to kind of dig into business intelligence and what is it, and how can companies use it, and Dave is an expert, so I’m excited to have him on. So let’s just jump right in, and Dave first maybe we could get a brief overview on your background and why BI? How you got into business intelligence and sure that will be great.

Dave DuVarney: Yeah, so I actually kind of fell into business intelligence, I graduated with an accounting degree, was doing work at a public accounting firm and realized that, that was not my bag, I didn’t like it, so I moved into, did some teaching for a while, I taught at a Microsoft Training Center for a year and then got hired at a newer consulting company that was a niche BI company out in Seattle, that was right around 2001, and that company there, they actually hired me as developer and I was supposed to do kind of custom development which I always love, still love, but they were a BI firm, so eventually I picked that stuff up and started doing BI projects and kind of been stuck with it ever since, so lots of fun, got to write a couple of books along the way and do some speaking around the world and so it’s been a good time being in that area now and I think I’m pushing 14, almost 15 years.

Dave: What were the books on?

Dave DuVarney: So, I did two books on Microsoft reporting services, so when it originally came out, before it came out I was doing these speaking gigs and a friend of mine said hey! We are going to write this book and I did two of them and the third one I passed, it’s a lot of work to do.

Dave: Oh yeah.

Dave DuVarney: But it was fun, it was great to get the name out and then as conferences came up, I had more opportunities to speak and go do those things, so yeah, it was a lot of fun.

Dave: Interesting, so those books are probably a little outdated by now.

Dave DuVarney: They’re a little old.

Dave: Yeah, a little bit old.

Dave DuVarney: The royalty checks trickled in, but they are not like they used to be.

Dave: Oh really?

Dave DuVarney: Yeah.

Dave: Can you still find one in Amazon?

Dave DuVarney: Oh yeah.

Dave: Okay. Would love to put those ____2:41____

Dave DuVarney: I’ve got one back ___2:45___ its even in Chinese which is pretty cool.

Dave: No way. We’ll have to put that one on the podcast today. So, tell us about Talvant, when you started it, why you started it, and kind of what you guys do?

Dave DuVarney: Yeah, so I had actually spent, I started my BI career in Seattle and I moved back to Madison about 6 years ago and continued to do work for my consulting firm and then was independent for a while and realized that I really wanted to build up a BI practice here in this area. I think there is a lot of talent in the area and I think there is a lot of opportunity to do it. The BI practices that I’ve seen in the area tend to have a couple of people, a few people and so what I really wanted to do is build a place where we had some of the kind of top talent and get those folks out and working and all under one roof, so we can share knowledge and do those things. I met my business partner Rob Long a couple of years ago, he was doing some external consulting along with his full-time job and the two of us just kind of said well, you know, maybe we should start something, so January of 2015, the two of us kicked it off and now we are going to be 7 full-time people come March, so pretty exciting and we are doing it all, it’s all our own kind of organic growth so we are getting there.

Dave: That’s exciting. We could talk a little bit more, I’m curious about, can you disclose some projects, but if you can’t that’s fine, but first let’s talk a little bit more about BI and what is it, it seems like a very broad, the term is used a lot, but it seems like you can cover a lot of different areas.

Dave DuVarney: I think it is and I think people get confused about different technologies and tools and there is really more to it than just the tools and the kind of IT pieces. If you boil it down from my perspective, it’s really saying, how do I use information to make an actionable decision right and whatever tool that is, whatever process that is, that’s what you’re really trying to get to. I have information, I am going to make a choice about it and I’m acting on that choice because of the information, not just because of my gut. To me that’s really more of what it is in a kind of a definition sense.

Dave: Gotcha. Okay. So in some ways is it similar to predicted analytics, I mean is that under the BI umbrella what you say?

Dave DuVarney: I think all those things fit under that umbrella right, so one of the things that people will look at when they talk about BI or data and making decisions is really, you’ve kind of got these perspectives that you look at so you can say what was, what is, and what could be. So when you talk about like analytics, you’re really saying kind of that what was, what is right. What was it historically, what is it right now, if I’m watching a manufacturing process, what is in right now and then when we talk predictive analytics, that’s when we are really saying what could it be, so we’re using that old data to inform what that item might be in the future and then we are going to make decisions based on that to kind of push forward, so really a time horizon of what you look at changes those ____6:05____ terms.

Dave: Okay, and are there different disciplines or areas within BI, like can you…
Dave DuVarney: Yeah, so we usually break it down, we talk about kind of tools right, experience and then we’re toying with the word “habits” right, that you have. So from a tools perspective and really architectural elements right, there is traditional data-warehousing right, where you’re going to get all the data in one place, and you’re going to structure it, and you’re going to format it, and that has been around for a long time and continues to move forward. You’ve got self service business intelligence, so all the new tools like Tableau and QlikView and Microsoft’s Power BI, where you’re empowering the individual to go and kind of build an analysis and make a decision, and then all of the big data pieces right, which are really kind of big up on the hype cycle right now and even kind of moving through the hype cycle really where you’re saying we’ve got these huge volumes of kind of a wide variety of data and now what can we do with it, so a little bit structured, but all those disciplines kind of fit under that umbrella.

Dave: Gotcha, and how do you know and may be give an example. How do you know what is this data-warehousing, kind of the Tableau of the worlds, yeah, how do you know…

Dave DuVarney: Where to go?

Dave: Yeah, where to go.

Dave DuVarney: So, there is a lot to that and I think that gets you in a kind of experience and habits parts of delivering BI solutions, but a lot of companies, what we find is that companies kind of go after a BI problem, they basically say we need to make better decisions with data and then they start automatically to like look at tools right, and they say ah! well let’s do big data because everybody is doing big data, so we need to do big data and what we push on is you need to take a step back and almost the tools, we can do the tools, the tools that’s not hard. The hard part is setting the priorities of what is it that you actually care about and what value is that going to bring back to your business and now let’s go figure out okay what tools because I’ve seen companies that will go down in Big Data Initiative and you’ll say well what are you trying to do? We are trying to improve our, you know, whatever our customer reach and so on, or do you have a CRM system yet? Well no we don’t have CRM or we have CRM, well how is your reporting on it there? Do you know how many opportunities you have and leads and the time it takes to close a lead? Well no. Okay, why don’t you figure that out first, so you know what’s in your pipeline and manage your pipeline, and then let’s go look at these other things that would better fill your pipeline right, so it’s just kind of this metric maturing through what you’re trying to look at and that’s I think where you get in and you got to help organizations to figure out your problem, the thing you’re trying, the value you’re trying to add. May be you do not need that tool, maybe it’s something simple. It could be ____9:09_____, it could be a simple self service thing or it could be big data.

Dave: How do you know when you’re talking to a company they say, hey! We want to try and do X. How do you know that the data given there that will deliver on what their expectations or what they want to do?

Dave DuVarney: Usually, that’s a big part of walking them through, so we’ll take them; if I started with a company from scratch, first thing we are going to do is identify what are their priorities. What problem are they trying to solve? What value are they trying to create? From there, then we are going to really go down that data path and say well this is great, so you want to measure, you know, again lets use the sales pipeline example; you want to measure how long it takes an opportunity to close. Do you have a CRM system; no we don’t. Okay well, we got to solve that problem first and then kind of move through those. So it’s kind of a discovery. Sometimes we’ll take people through this exercise of, and we have some templates we have used over the years that will help them identify all the key metrics they want to measure and then we actually take them through a scoring of technically how feasible is that. Do they have the data, don’t they have the data, and then we basically plot it out and we’ll get a little four square that size that says okay well based on, yeah, this one’s really important but you don’t have the data so you can’t do it, you got to do this one up here until you fill that gap.

Dave: Gotcha. What goes into that scoring model?

Dave DuVarney: So it’s just an old tool that I’ve had a while and on a business side it takes into consideration things like, how many users are looking at a metric, does the metric drive a strategic objective, does the metric drive profitability which those two things can be slightly different and there is two other kind of business sides, so we get 5 things we look at from the business, then on the technical side is the data available or not, how hard is it to get it, how clean is it, all of those pieces and that kind getting into those two axis plot against each other to give us what we should go after first.

Dave: Well, that’s a smart way to do it. I didn’t think about that. Yeah, it probably helps like you said make a go or no go decision and what to focus on. Interesting. So how many different areas would you look at, like on, you know, this chart there or graph, how many plots would you have?

Dave DuVarney: It’s usually kind of like a 5 to 10, somewhere in that neighborhood, so we take, you know we could sit through, if we sit down with the business, and where that all starts is we say what are the questions you’re asking about your business right, and if we do this kind of in a full context, we’ll actually talk to directors and VPs of each major business area and see, first of all what are your goals and objectives? What are you trying to do in your area? What’s the information you need to get to those? So we start the conversation there and then we basically drive out a bunch of business questions and those are the things that give us all these little KPIs and the KPIs get grouped together in big buckets which are usually sales, marketing, customer service, I mean they fit kind of major business areas as you go.

Dave: Interesting, okay, and so once you answer some of these questions and kind of figure out their vision, how do you actually implement this? What tools do you use? I know it probably differs depending on the project, but can you give one or two examples of how you actually implement this?

Dave DuVarney: Yeah, so from a technical implementation standpoint, our strength is on the Microsoft stack because that’s what we have been around the longest so for us, you know, we start to look at what are the ETL components that we need to build, so how we are going to move the data out of the core systems and centralize it somewhere and pull it together, because we do a lot of that traditional data-warehousing piece. How we are going to extract that data, what frequency, what’s it going to look like, then we get into modeling which is another big piece and kind of, if you talk about big data, that’s where these things start to diverge a little bit, but we really look at back to the sales pipeline example. There is a way to model that data, so you can understand opportunities over time and so we do the modeling, do detail work to pull it into a central area and then we will build the semantic layers again with Microsoft Analysis Services and things on top.

Dave: And what you mean by modeling? What would that look like?

Dave DuVarney: So when we get into it, so the data modeling piece I think is pretty fundamental to having some success in actually reporting the information. There are kind of schools of thought on what types of modeling are best so a couple of the big ones, there is Kimball data modeling which is, you know, people will refer to it as dimensional modeling also. There is Inmon, which is a little more of a normalized type of modeling, you have traditional just third normal form database modeling and then there is another one that we have been exploring lately called Data Vault, and really what it is, is just saying you need to transform the data coming out of source systems in general is not the best for reporting, so any of these techniques it’s really, you’re looking at what type of reporting are you trying to get, and then you’re going to go and re-shape the data to match that to get out the other side.

Dave: Gotcha, makes sense. So you use the Microsoft’s Power BI. What do those tools look like, is it sweeter tools, how’s it?

Dave DuVarney: So the Power BI piece is kind of interesting. I mean it’s evolving quickly and I think along those same lines Tableau and QlikView fit into that what we call that self service BI area right, they are really intended for the technical user, but not super deep technical user, and they let you do things. Anytime you are going to go and get data you’re doing ETL. You’re going to extract it, you’re going to transform it, and you’re going to load it into something, and so that’s what those tools are doing, like Power BI is giving you all these connecters into QuickBooks and relational databases and everything else, you get it out, then they have got this thing, Power Query on top that allows you to transform it, so let’s say you got somebody’s first and last name and you want to split it apart, you have these little transform expressions to kind of cleanse it, manipulate it, and then the load goes away and then you are really just visualizing it, so you have got all these expressions to go in and visualize and pull those pieces together.

Dave: Gotcha. Okay. So now that we have all the data together from different systems, it could be like the CRM system, Accounting system wherever you need the data in order to give those intelligent insights, you know, keeping all the data together which is like you said complex. So let’s say the data is together, now what? You have all this data, you probably already thought about the questions you want to ask of the data, is that right? And then so how do you actually start asking these questions and kind of empower the users?

Dave DuVarney: I think that gets into really a lot of the process elements around, we’ll go talk to people and they’ll say yeah, we want to do business intelligence or we want to do data-warehousing and they look at it like a project right, like oh! I got the data here and there it is and you have delivered this thing, now go. That’s not really how it works, and I think that’s where BI Solutions are kind of unique from like a CRM system. If I put a CRM system in front of somebody and I say here’s CRM, you’re going to track sales leads, that’s the system these people have to use to do their job, like they have to put their leads in there, they have to do that, they’ll have to, but it’s going to help a lot right. A BI system, I think you have to go evangelize. So you want to understand what the business wants and then you have to have business analysts or people who are supporting or building that solution, who go and evangelize it and say look this is what you can get if you come here and use this piece and so when you have a really good BI program, you are going to have somebody that’s that front interface to the business, that’s going out, talking to them and saying here is this, and they are both going to mature over time, so the business is going to get smarter about what data they have and now what could they do next and you have this kind of constant interaction cycle that keeps a BI program kind of going on as long as the business goes on.

Dave: Makes sense. So can the users ask questions? Are all the questions formatted ahead of time or can they ask more open-ended questions?
Dave DuVarney: Or form new ones?

Dave: Yeah, can they form new ones?

Dave DuVarney: I think that’s where you get into some architectural elements around, you know, how do you expose the data, so traditional tools like cubes and data warehouses, you really have to find for the user what are the metrics they can have…

Dave: What are cubes?

Dave DuVarney: Cubes are multi-dimensional databases that just allow me, basically if you ever worked with an excel pivot table, the backend can be the backend for a pivot table, so you can go in and say, I’d like sales and cost of goods sold, and I’d like to see a byproduct for last year right, and you drag all these elements together and you have got an answer. Those pieces were very structured through IT right. You had to build them and put those things together. What we’re seeing now is that, that will answer some of the questions and you want to have that there I think for the users, but the pieces where they want to do their own exploration, that’s again where Power BI and Tableau and these QlikView pieces come in because you expose them kind of further back in that data transformation chain, and you say okay here’s the data in a little more of a raw form. You go point those tools at it and you basically build your own model and get your own analysis out of it, and then if they do something really well and it looks great like somebody in a department that’s a good analyst, they build something that looks nice and their department wants to use it and the rest of the enterprise wants to use it, then you back that back into the architecture and say, okay now let’s go build that. Let’s take your model that lives in this little workbook and let’s go build that back into the enterprise, so anybody who connects to it with these visualization tools can just kind of drag and drop.

Dave: Until all that exploration, that is probably what I’ve thought needed around the initial architecture in order to even make them happen.

Dave DuVarney: There is a lot of thought around the architecture of how the data moves, so you know when they’ve landed where because inevitably you’re going to have to go figure out why something is not what people think it is and go back, and then that gets back to the modeling piece, how you model the data to expose it to the users makes a ton of difference into what their experience is going to be trying to consume it. Most of the times when I get into something where somebody’s built a solution that people don’t like or have not adopted it’s because the modeling behind it was terrible, not because the ideas were wrong or anything, it’s just that the modeling, somebody didn’t do the modeling right and then everything just doesn’t fit together like it should.

Dave: Do you have an example of that off the top of your head? It’s fine if you don’t.

Dave DuVarney: Yeah. Not specifically.

Dave: No that’s fine. Yeah, understandable, you’re gone. So speaking of an example, I’m curious if you have like a case study or something that you can walk us through. We have lots of interesting ideas, but for me, I always like to hear consulting.

Dave DuVarney: Sure.

Dave: How does this actually work? Like from the beginning to the end, how does the project work? You don’t have to name names or you know…

Dave DuVarney: Well. So we’re working, you know, Talavant is pretty new. So of our clients, our first client is a manufacturer up in Northern Wisconsin and they do customizations around trucks right, for fire departments and other things and basically what we came into their scenario is they had a bunch of excel spreadsheets and a bunch of things that different accountants throughout the organization put together and it would take them several days at the close of the end of any month to actually produce something back to the ownership of the business right, so they had to get all the spreadsheets together, bunch of people had to work on them, then they had to collect them all into one place and then give them to the owner so the owner new what’s going on 7 days after the month is done right. Now if you’re a business owner knowing 7 days after your month is done that you had a terrible month, it’s helpful, but it’s not as helpful as if you know as the month was going what are those inputs that are not going to happen.

Dave: While those spreadsheets are buried in expenses, productivity, and maybe a pipeline of the future, does it cover…

Dave DuVarney: Yeah, it was covering orders, so from their perspective, they were covering a lot of stuff around open orders, so how many open orders do we have, you know how many are moving through, you know what’s that production cycle look like from the customer wants it until I’ve put it out the door and I’m going to collect some money for it right, and you know like any business, you are going to want to tighten that up as much as you can and you are going to want to have insights into when there is hiccups in it and you don’t want to have that a week or two weeks after something has happened or in this case it would be like it could be 6 weeks right. So we went there, we started pulling their data into a centralized location, modeling it so that they could do analysis. We started out with their financials, so we built them a general ledger cube, so they could go in and just basically pull any kind of general ledger report they wanted to through excel and just pivot and twist and turn and do these, you know, some people kind of if we look like big data and those things are happening now, that’s kind of an old school approach, it was a huge benefit for them.

Dave: Wow. What about like the initial questions? I mean it sounds like they kind of already knew what they wanted. I bet companies often know what they kind of want, but how do you make sure it’s actually what they need?

Dave DuVarney: What they need?

Dave: Yeah.

Dave DuVarney: So for this one, we kind of had a direction already set when we came in right. They were kind of already set, and we’ve had another one here in town where we went through the actual kind of road mapping exercises and that was an 8-week exercise where we went and this was a bigger company and we actually talked to the directors of every single business group, we must have had 20 plus interviews and going through, and what you’re trying to do, where you’re trying to go, and from that we actually lay out a roadmap and we said, you know, first thing we need to tackle is sales, second thing you need to tackle is supply chain, the third thing you need to tackle is finance, and then started the layout of work to actually go down and tackle those things.

Dave: Did you use the scoring model for that?

Dave DuVarney: We did on that one in different pieces. Yeah, because we had all the questions at that point and we could kind of ask them okay, how hard or how easy is this to carry out?

Dave: So you get all the data together, and how long did it take?

Dave DuVarney: It is continuous.

Dave: Continuous, okay, yeah and that’s BI like you said, it’s just a continuous process.

Dave DuVarney: Yeah. I mean when we define, those areas, those big buckets, those we call them “subject areas” right, like sales is a subject area. Our subject area generally to do the first route takes 2 or 3 months to get everything and they go quicker as you go, and then what you find is after a year or two years, you’re going to want to go back to that because again the business has gotten mature or changed or whatever it is and you kind of come back and you redo some of those.

Dave: Gotcha. This is kind of a side question, but how do you, if a client is like well how are you going to measure ROI and just like how do we know if it makes any difference and maybe it’s enough to say hey we are going to get reports the day after it closed and that’s, maybe they don’t need an ROI, but does the ROI come up much?

Dave DuVarney: It’s tricky. It is tricky. I do have one that we were working on, we are actually working on it right now where it was easier because they had a very defined problem around reliability and with that one they could a put a number on it and so then it was easy to say, well here’s the number of what you’re missing and here’s what the program is going to cost, and then that was a no-brainer, but I would say that was very much outside the norm, very much.

Dave: It’s hard to come up with ROI typically.

Dave DuVarney: It is tough because sometimes you get into it, I mean sometimes you can do it because you can just say, well if you have visibility into X, how would that change your margin? We can get another percentage of margin, okay great, now we have a number right, now we can put a dollar figure on it, but sometimes that gets tricky or the other one we get into is like cost, what we spend X amount of hours a month doing this, that one usually is not quite justification.

Dave: And with the manufacturing example, after you have allowed the users to run their own reports, do they come back with any feedbacks saying how this really changed this or maybe it just changed our practice this way. Did you get any feedback?

Dave DuVarney: Yeah.

Dave: Did that happen all the time?

Dave DuVarney: Oh yeah, yeah, I mean if you’re doing it well that’s what you are going to get and that gets you into that piece of being that analyst constantly interacting with them to say okay, are we on the right path, are we doing the right thing, and bopping around and can’t get enough feedback so…

Dave: So switching course now to the future, where you think BI, well this is a big turn, but where is BI headed? I mean, it sounds like there are a lot of firms that are just doing, I don’t want to say traditional, but just they are still a lot of work to be done just to get them up to speed essentially.

Dave DuVarney: A ton.

Dave: Yeah.

Dave DuVarney: I think the tools are improving right, so you’re getting more of an ability to go after pieces of information that you couldn’t get before, and if you look at things like what’s going on with Big Data right. All of a sudden you can get terabytes or petabytes of information that you just couldn’t get, a lot of it is unstructured and so what I think is going to happen in the longer term is this notion of kind of semantic modeling and understanding what the data means will kind of come full circle again. I think it’s really cyclical right. Like ETL tools weren’t fantastic a long time ago and now it’s easy to build, not easy, but easier to build ETL pieces, getting some of these big data pieces is tough to pull those things together and to do the analysis on it and what I think will come next is usually we will say, well now we need to have governance around it and know our quality, and then we need to get a semantic layer on so we can interact with it, I think those are the pieces that will kind of come down the road.

Dave: More incorporating the Big Data piece.

Dave DuVarney: Yeah, I think so. I mean it will just become part of the mainstream, but I think we’re going to always come back to some context of what does this data even mean and what problem is it helping me solve. I do think the other thing that is out there that probably hasn’t been as prevalent yet is the usage of cloud technologies just to kind of get these things going. A lot of times you start like a big BI or data warehousing initiative. If you have serious volumes of data, you’re buying some pretty beefy hardware to deal with this stuff, so you look at these initial investments that are in the $10,000 to $100,000s of hardware to just get going. I think the cloud pieces offer you this kind of cheaper entry point where you can say well, I’m not sure if this is going to pan out to be something big, so why don’t we start with, you know, let’s move this stuff into Azure with ____30:11____ Amazon and then let’s build from there and see how it goes and then maybe we bring it back on to parameter or wherever it is.

Dave: Yes, it makes sense. Okay, and we are almost out of time here, but this has been great. So if people want to learn more about BI, do you know of any BI kind of educational resources or any place people can look besides you?

Dave DuVarney: Gosh, well, information management is a pretty good site that covers a broad set of topics, so I get those e-mails every day to kind of help me go through. I think people that are looking to practice in this area, I still put people back to you Kimball’s Data Warehousing Lifecycle Toolkit, and it’s an older book, it does get you into a lot of the dimensional modeling concepts, but it also talks about the prioritization process in really helping, you know, what’s the process to take the business through to help them define the right metric because people miss that way too often. I don’t take them through what’s valuable to them. You need to do that first.

Dave: Interesting, and it sounds like it would be helpful to have a technical and a business background for somebody who wants to get into BI.

Dave DuVarney: Yeah, ideally.

Dave: Yeah, may be somebody that is technical they can learn the business or part of it too.

Dave DuVarney: The people I’ve had the most luck with, I used to do a lot of recruiting folks out of college and I actually had a lot of luck with people who came out of Econ backgrounds, and any kind of engineering, it can be mechanical, electrical, or whatever it was. They just were weird to kind of do that and music.

Dave: Of course, right.

Dave DuVarney: Some of the best people I know in his area studied music.

Dave: Really?

Dave DuVarney: Hmmm.

Dave: Do they have to be that technical to get into this space?

Dave DuVarney: You have to have aptitude right. I mean, I think some of the like hardcore computer science stuff not so much, but you have to have an aptitude right around the computer.

Dave: Right. Like aptitude and learn the tools and just be curious.

Dave DuVarney: Hmmm.

Dave: Yeah okay. Alright, well, this is awesome. I think that’s all we have for today unfortunately, because I could ask some more questions, but I definitely appreciate your time Dave.

Dave DuVarney: Yeah.

Dave: Yeah, until next time.

Dave DuVarney: Alright, thanks for the opportunity.

Dave: Thanks everyone for listening. We’ll see you next time.

Dave DuVarney: Bye.