SDS 049: Great Tips on Building a Successful Analytics Culture - SuperDataScience - Big Data | Analytics Careers | Mentors | Success

SDS 049: Great Tips on Building a Successful Analytics Culture

Welcome to episode #049 of the Super Data Science Podcast. Here we go!

Today's guest is President and Founder of Tiber Solutions Jim Hadley

Subscribe on iTunes, Stitcher Radio or TuneIn

Bucking the trend of many conventional consultancies having highly specialized data consultants and analysts, Tiber Solutions goes for the “Swiss Army Knife” approach, to great success.

Tiber Solutions has just been named one of the Top 50 Places to Work in Washington DC and its Founder, Jim Hadley, joins us for today’s episode.

If you’ve ever wanted to know what goes on behind the scenes of an analytics consulting firm, how to get hired by one of the top analytics consulting firms in Washington DC, or how to build a successful analytics culture, tune in now!

In this episode you will learn:

  • Being a Data “Swiss Army Knife” (07:34)
  • Reasons for Having Multi-Skilled Consultants in Analytics and Business Intelligence (14:01)
  • A Different Approach to Hiring and Building an Analytics Culture (23:16)
  • Knowledge, Understanding and Wisdom (31:50)
  • Case Study: Bringing Self-Service Analytics to an Organisation (33:50)
  • Overcoming Organizational Obstacles in Data Culture (39:54)
  • Star Schema Deep-Dive (47:25)
  • ETL Deep-Dive (51:02)
  • When NOT to use a Star Schema (52:51)
  • Tools vs. Solutions (55:36)

Items mentioned in this podcast:

Follow Jim

Episode Transcript

0

Full Podcast Transcript

Expand to view full transcript

Kirill: This is episode number 49 with Founder of Tiber Solutions, Jim Hadley.

(background music plays)

Welcome to the SuperDataScience podcast. My name is Kirill Eremenko, data science coach and lifestyle entrepreneur. And each week we bring you inspiring people and ideas to help you build your successful career in data science. Thanks for being here today and now let’s make the complex simple.

(background music plays)

Hello everybody and welcome to the SuperDataScience podcast. Super excited to have you on the show, and today we've got an exciting guest, Jim Hadley, who is the President and Founder of Tiber Solutions. Now Tiber Solutions is a consulting firm specialising in analytics consulting. And today's podcast is going to be very different to what we're normally used to. So normally we have quite a lot of technical stuff on the podcast, so we talk about some tools and techniques and methodologies. And also we sometimes talk about how people build their careers, or how people have built their careers and how they've progressed throughout their lives in the space of data science and analytics.

Well today, we're talking about the other side of the coin. We're talking about what it is like to actually run an analytics company. What it is like to build a culture, to maintain a culture where people are advocates of data and at the same time, they're happy in their workplace, they're enjoying their work, they're doing the things they love.

And this podcast is going to be useful to you regardless of whether you're a business owner or an executive or if you're looking for a job. Obviously if you're a business owner or executive, this podcast is going to be useful because you will get some tips and hacks on how to create that analytics culture, that environment to foster talent and help people flourish and learn the skills that they want to learn. On the other hand, if you're looking for a job, this podcast will actually show you what a successful analytics workplace should look like. So what you should be looking for. Because a lot of the time, and I am not tired of repeating this, a lot of the time, people get it wrong. People go to interviews and they just want to sell themselves only, they just want to just get the job at whatever cost. But at the same time, the interview is as much for the employer to interview you as it is the other way round, for you to interview the employer to understand what you're getting into yourself.
And this podcast, thanks to Jim, will be able to discuss that, investigate that further. Because Jim has been successful in building a great analytics culture. His company, Tiber Solutions, has been named one of the Top 50 Places to Work in Washington DC. Just recently they got that win, and that's a big thing for a small analytics firm to be one of the top 50 companies to work in in Washington DC, that's a big deal. And of course, we're going to dig deeper into that. So if you want to find out more about what kind of culture you should be looking for in an organisation where you're going for an interview, then this is the podcast for you. I can't wait to get started, and without further ado, I bring to you Jim Hadley, the President and Founder of Tiber Solutions.

(background music plays)

Welcome everybody to the SuperDataScience podcast. Today have the President and Founder of Tiber Solutions, Jim Hadley, on the line. Jim, welcome. How are you today?

Jim: Doing great, Kirill. Good to be with you, just looking forward to this time.

Kirill: Awesome. Thank you so much for coming on the show. And just for our listeners, where are you calling in from?

Jim: I am calling in from the suburbs of Washington DC in Virginia.

Kirill: Fantastic. And how's the weather today?

Jim: It is beautiful. It is cherry blossoms. This is the couple weeks during the year that the cherry blossoms are out and it's 70 degrees Fahrenheit and just beautiful.

Kirill: That's so cool. What is 70 degrees Fahrenheit in Celsius?

Jim: Oh man, that's a hard data science question. I have no idea. It's shorts weather. It's just starting to get into shorts weather is all I can say.

Kirill: Oh my god, yeah, I always get confused with these things myself, but it sounds pretty cool. I'm already jealous because here in Brisbane, it was raining crazy last week. They had to close down the whole city on Friday, it was so much rain. We’re going through a bit of a wet season right now. Yours sounds a bit better.

Jim: Yeah, you guys are going in the wrong direction, heading into winter.

Kirill: Yeah. Oh, well, it happens, I guess. You’ve got to have some winter to have some summer in the end. It makes you appreciate those things more, right?

Jim: That is true. Very true.

Kirill: All right. Jim, I’m very excited to have you on the show. You are the President and Founder of Tiber Solutions, which is a data science and analytics consulting firm. Can you tell us a bit more about what does your company do and how long have you guys been around for?

Jim: Yeah, we do just one thing, and that’s analytics. And that is every possible area of analytics from just traditional data warehousing and business intelligence and ETL to pretty sophisticated data science type stuff as well. We’ve been around since 2005, is when I founded the company, but I’ve been a practitioner since about 1994, so about 23 years.

Kirill: Wow! Okay, so you’ve definitely had a lot of experience by the time you founded the company.

Jim: Right.

Kirill: And what areas of analytics were you into yourself before starting the company? Were you just database-focused or visualization-focused, or did you have kind of a broad spectrum of everything?

Jim: I had a pretty broad spectrum for that timeframe. I started out at Accenture in 1990 and was there for about 4 years and then about 4 years in I got put on my first business intelligence project, it was for a regional healthcare company, and I continued to work there for about 4 years.

Right around the mid-90s, Ralph Kimball came out with his first book on star schema, or dimensional modelling, and I just fell in love with that. This is a very elegant, beautiful way. It’s brilliant to actually model data in a relational database that’s high performance but just being able to really correlate information very flexibly and intuitively.

So I really just started being a practitioner of that and after a couple of years, so about 1998, I looked at Ralph Kimball’s website and—you know, my wife and I were starting to have kids and I wanted to get off the road and get out of the Big Four. I looked at Ralph Kimball’s website and he recommended three people in the world that were experts in star schema. One of them was a guy named Chris Adamson that was basically in my backyard. I gave him a call, he was about my age, and we got along really well. He didn’t own this company, but he was working for a small boutique business intelligence data warehouse company and he and I became the senior consultants there.

I went on to a couple other start-ups. Ralph Kimball was heavily involved in this company, so really during that timeframe I was mentored by the industry leaders in star schema and just had a great opportunity to play many, many roles. I was the ETL developer, the star schema designer, an architect, the DBA, the BI developer. I did all the designer requirements. I was able to go in and talk to a CIO about strategies and tools and methodologies and best practices and build the whole system out for him or her, but I was also able to cross over to the business side and speak with the Chief Marketing Officer or a CFO and really listen and understand their business and understand their business problems and be able to help them understand the enabling capabilities of these products and these tools and co-invent a solution with them, touching base with them multiple times through the week and showing what I was doing and maturing it and getting their buy-in and their direction as I went.
So that was kind of my background. I played every role there was in more traditional business intelligence and data warehousing. In 2005, I kind of had a pause in my career and that was kind of the reason why Tiber was started, was because I had seen that that was a very successful way, to kind of be a Swiss Army knife, to be deep and wide in this space, and that was the best way to create these solutions. Because there are so many design decisions that need to be made, and unless you have experiential hands-on knowledge of all the different layers and the components, it’s hard to make a wise choice in how to implement the requirement.

At that time in 2005, I just saw that the Big Four and a lot of companies were just putting 20-30 people on these types of projects and everybody was so specialized and so typecast into their responsibility on that project that there was nobody really that saw how those puzzle pieces needed to come together. Yes, there was a project manager, but they had no hands-on experience in those and they were just making sure things got done on time and on budget.

So I wanted to be that brain. And that’s really what Tiber is, it’s a company that is small intentionally because we don’t really hire junior level people. Every person at Tiber either is or aspires to be kind of this Swiss Army knife that can do it all. And that’s how we think projects are successfully done. So instead of putting 20 people on a project, we’ll put like 4 people on a project and these people will share all the responsibilities and be able to have wisdom in getting the work done.

Kirill: Yeah, totally. I can attest to that because a couple of people from your company – I think Jeremy and a few others – have actually done several of my courses. It’s very interesting to see how you actually say that you’re creating these Swiss army knives out of people and that they’re not just focused on one thing, they’re focused on many different things and they’re building up their capabilities in all these different areas. I would imagine that that is a much more fulfilling job, but what are your clients saying about that? Are they impressed or is that not what they’re expecting when you guys come in and you have 4 people instead of 20 people and they manage to get the whole job done anyway?

Jim: It’s a very different model than what they’re used to. Sometimes they get it pretty quickly, sometimes it’s a wait and see and sometimes they just continue to scratch their head. They’re used to having more of a waterfall approach at times and having 20 people. They’ll look at me and say, “Where is your business analyst? Where is your DBA? Where is your data modeller? Where is your Tableau developer? Where is your SAS developer?” and it’s like, “That’s Jeremy. He’s all of those.” And they’re like, “Well, how can that be?”

It’s like, “Well, we’ve trained them and we’ve mentored them and we’ve apprenticed them and we’ve been very intentional about that.” It’s just really a part of our culture. So instead of having 20 people, we’re going to give you 3 or 4 and they’re going to be interchangeable in a lot of ways. And instead of having one person thinking about this star schema, you have 4 people thinking about the star schema and there’s wisdom in numbers there. They know all the different layers, all the different components and how all these pieces need to come together and they can weigh all of that because they’ve played these roles multiple times and they’ve learned through experience.

I think sometimes it’s just a strange model for them to get their head around. They don’t really trust them, like “How can somebody really know all of that? It’s like saying somebody is the world’s greatest Java developer and the world’s greatest Oracle DBA and the world’s greatest business analyst all in one.” Those guys don’t grow on trees. You’re right, you have to be very intentional about it and that person needs to want to be that type of person and they’ve got to have the DNA to have that as well. You know, in some cases they like the idea. They’ve seen large projects where a baton gets dropped as it goes through the gauntlet from requirements to final testing and production releases. They’re seeing less people that they’re paying billing rates for, too.

Kirill: Yeah, totally. That’s a very interesting concept. For instance, because we have quite a few listeners on this podcast who are either business owners or entrepreneurs or, even more so, we have quite a few people who are in managerial or executive positions in companies. Do you think it’s possible to build a team like that, not as a consulting team but as a team inside your own organization? Or do you have to pigeonhole people inside an organization and make sure that everybody is just very good at what they do but they’re only doing one thing? Or can a person inside a company, like an executive, strive to build a team of data scientists or analysts similar to what you have in your consulting firm?

Jim: Yeah. I personally think this is the best way to create quality solutions. Analytics and business intelligence, it doesn’t work well in a factory assembly line. Things get lost throughout the process. You need to kind of see the very beginning, how the source system is collecting the data and what the business rules mean there and bring it all the way through the process and kind of begin with the end in mind, to the actual problems and the analytics that people are going to need to solve those problems.

So I would definitely encourage companies to hire employees that want to be that way. It’s fulfilling for employees. We do one thing, and that’s analytics. I’m never going to hire a Java developer, or a SharePoint administrator, or for that matter, a project manager. These guys love the fact, and I think everybody would, to continue to learn and to be challenged and to be given opportunities to grow and never get old in their skillset.

I think it’s a little bit harder for some of the consulting companies because the model that we have kind of takes away from their desire to put as many people as possible on projects and kind of justify that number of a larger project team. So I think that’s something a little bit different about our company, too. We believe that quality is the most important thing, and we have a very profitable company. We don’t have tons of overhead, we have no office space. The two big expenses we have are salaries – we try to pay very, very well – and healthcare expenses. I think we’re one of the last companies in the world, or at least in the U.S., that pays 100% for our employees’ and their families’ healthcare. So employees have no cost for all their health benefits, and that can be significant. I mean, in the U.S., for a family, that can be $25,000-$30,000 a year right there.

Our model is more of “Let’s keep our company profitable but let’s focus on quality, and when the quality is there, then revenue comes and additional work comes,” as opposed to trying to jam as many people as we can on a project and trying to justify that.

Kirill: That’s a great approach. I’d like to dig in a bit more into that, how you actually manage to create a company like that, because I think for all our listeners out there, this is going to be a good example of not just how to build a company, but also what to look out for. If you’re looking for a job, then what do you look out for to make sure you end up in an organization like this one, an organization where you’re valued, you have amazing opportunities for growth, you’re not just being pigeonholed into one specific task which won’t allow you to develop further in your career in the directions you want to, plus you’re very well-looked-after by the company. I know that this is true about your organization because you guys were named the top 50 places to work in Washington, D.C., so that’s a huge accomplishment for a small company who are in the consulting space of analytics. Congratulations on that, by the way.

Jim: Thank you. Whenever someone brings that up, I’m almost moved to tears, to be honest, and it’s just out of humility. There are a lot of awards in D.C. for companies, and you can kind of game the system. And a lot of it is based on revenue or year over year profit or growth. Frankly, those companies could be driving their employees like slaves and still get those awards. This award has nothing to do with revenue. It is just about employee feedback.

So, greater than 75% or so, a large section of the company needed to submit pretty detailed surveys. And it was just based on what employees thought of the company, and the fact that our employees feel that way about our company—it’s really their company, right? The culture that they have created and they contribute to every single day is humbling to me. It’s humbling to me. I’m not that smart, I’m really not. We’ve got some really great, smart people and a very good learning environment where everybody knows that there’s somebody else that knows something more than they do in some area, and it kind of keeps us humble and wanting to be kind of lifelong learners and always sharing and contributing.

Going back to your question about how I wanted to start this, I guess I just looked at the first 10 years of my career where I worked in a Big Four company for eight years. The typical thing with Big Fours is it’s great for some time, but at some point you get sick of the traveling, you usually get – at least in the U.S. – kind of pigeonholed into one area of technology and they want you to get better and better in that one area so your billing rates can go higher and higher. And after about four years or so, then they want to switch you over to be more of a project manager and you become less technical and you don’t know exactly what’s going on anymore. You’re trusting people below you, you’re getting more and more removed and you don’t have that wisdom that you can share.

And as you keep going up, as you’re a partner, it really gets down to you’re a sales guy, how many people report to you. Maybe it took me a while, but I thought that was my definition of success – and maybe this would be a crude term – but I think that’s really just management. That’s management.

At the same time, when I was leaving Accenture, I was looking at Ralph Kimball. I just fell in love with this guy, I was like a disciple of him. I was like, “This guy has two people in his company – him and his wife – and he’s probably working for his wife, in some ways.” But yet, he would go in to organizations that are some of the biggest organizations in the world with the craziest, hardest data problems and analytic problems and solve their problems. And then on their off weeks, he’s having these 200-300 room training sessions where he’s teaching literally a generation of practitioners how to use star schema and implement that in an elegant way.

I was like, “That’s my definition of success. He’s a leader. He has nobody reporting to him but he is a leader.” I realized that I really wanted to just try to be the guy who knows his stuff really, really well and have a company that’s based on wisdom. All of these things kind of came together where we have a company where we don’t travel that much and we don’t work crazy hours. A lot of times when companies are working crazy hours it’s because some project manager that doesn’t really have any hands-on experience created some project plan. In our projects, we create that. I don’t want to have to pay for the sins of some project manager that really doesn’t know this stuff.

And then I was being typecast, I was being removed out of the technologies, so I wanted to create an organization where people could stay in the technology and embrace that and be rewarded for that and be leaders. We don’t have titles in our company. We’re a really horizontal company. It’s like myself and our CTO Greg Jones and that’s kind of it. Everybody else is somewhat at the same level below us. I don’t know if that answers your question, but those are some of the ideas that would kind of float around in my mind when Tiber started.

Kirill: That’s very interesting. Is that the first thing you focused on when you started Tiber? Do you focus on culture? What were the first couple of things that you looked at? For instance, why I’m asking this question is if somebody is listening to you and is hopefully getting inspired by the culture that you’ve been able to create, the analytics advocacy culture, and they want to replicate your success, what would you say the first thing to focus on, or the most important thing to focus on, is for somebody in that position?

Jim: It’s definitely the people. A lot of companies say this. A lot of companies say that employees are their greatest assets, but for us it is absolutely true because frankly, it’s almost impossible to find a person that is as deep and broad in analytics as the folks in our company. The world doesn’t give people those opportunities. Like we said, in a Big Four consulting company you are going to be typecast and you’re not going to have opportunities to grow.

And the folks that do and have been exposed to a lot of these technologies and best practices and skills, you know, they’re probably working for a small company, or maybe directly for a CIO, and the CIO is like, “Oh, just go figure this out.” So they’re not properly mentored. We look for people that are not straight out of school or even probably have a good 3, 4, 5 years in a consultancy, have mastered one or two of these areas and have just a real desire to continue to learn and value being mentored and are just hungry. And they have the DNA for this.

Those are the ideal people that we’re looking for and then we can bring them in and begin to train them and mentor them and then put them on projects and opportunities where they can kind of be paired and cross-train with a person. So someone who is fantastic at Informatica is going to be paired with someone who is fantastic at Tableau. You know, they’re going to kind of cross-train each other as they’re working through on opportunities and projects. So it really is the people that make this possible. You can’t just hire an Informatica guy or a SAS guy or a Hadoop guy and say, “All right, they’re just going to naturally blossom into a well-deep/wide Swiss army knife skilled person.” So you kind of need one person that can kind of be that head and that mentor and that Ralph Kimball, and then start bringing in folks that have that desire that can be mentored by those people.

Kirill: Gotcha. And where do you find these people? Like, these people sound pretty rare based on your description. Where do you look for these people?

Jim: Probably the best place that I look—I’ve just gotten to be really good at looking at somebody’s LinkedIn profile, believe it or not. Those are always the best people, because they’re usually not looking for jobs but you can look at their background and their passion and what they’ve done. So I usually will just start reaching out to some of them and describe our company a little bit. And it’s different, right? And people are like, “Well, I’m not really looking, but you sound kind of different. I really want to know a little bit more.” So I have that conversation with them and just plant that seed. And sometimes people are ready to move and we continue the process.

We have a pretty good recruiting process where I usually talk to them first and then I have our Chief Technology Officer meet with them face-to-face and then I have them talk to at least two of our Tiber guys. And I encourage our prospects to ask the hardest possible questions, the most difficult questions that would embarrass Jim. You know, “Jim’s been living in this Tiber bubble for 12 years. He talks a pretty good game, but do his words translate into actions in reality?” I just want to make sure that anybody who comes to us, there are no surprises. They know exactly what they’re getting into, so that’s good.
And we also do what we call Tiber training. We actually had a Tiber training session last Friday. We do it every other month on a Friday and we’ll just shut the company down for a half day, go into a room and have a whole company there and we have a couple people present or facilitate some of the hardest problems that they’ve seen in the last couple of months.

Sometimes that’s new technologies, more hands-on, everybody’s logging into AWS and doing something, and sometimes it’s little more softer skills like, “I had a customer and this is what was happening.” It’s not about giving them the answer, it’s more about—these are the things that you can’t usually read in blogs or books. We’re trying to share wisdom, so there’s not really a black or white answer. There’s good and better, but let’s kind of explore that together.

So we usually have our prospect employees, our candidates come join us. We had like five of them, which is higher than we usually have, but we want to see how they interact with everybody and see how they fit into the culture, the chemistry that they’re contributing during the learning session and just if it feels right. We do happy hour afterwards as well. It’s been really good to just build corporate culture because everybody is coming in from different projects and sharing 6-7 times a year, but also to help people that are thinking about joining Tiber get a preview of Tiber in a microcosm.

Kirill: I love it. I especially like the example which you gave where you talk to a person on LinkedIn and then kind of plant the seed in their head, even way before they’re possibly ready to join or leave their company. So you’re looking out for talent. I think a lot of companies make the mistake that they only start looking for somebody when they desperately need them, when they’ve identified the need to fill a role and then they want to fill it right away. And that way they’re looking for whoever is available or whoever is looking for jobs, and that way they might fall into a trap of hiring a person who is not the optimal person for that role, whereas you’re just browsing LinkedIn, I can imagine you sitting with a cup of coffee in the evening just browsing LinkedIn on your iPad and you’re like, “Okay, this person looks interesting. How about I chat with them?” And then one and a half years later they happen to join your company because you had that chat and you planted that seed in their head before they were even ready to leave their existing employer. I think that’s great advice for those who are looking for—

Jim: Absolutely. We call it almost ‘dating’ – I’m doing air quotes here – because it kind of feels like dating, just “Will this work out on both people’s end?” and we need to make sure that it does. Sometimes that takes a month or two and there’s an opportunity—you know, sometimes we’ll have an opportunity where this person clearly is qualified for this opportunity but they’re just not interested in it. That’s fine. We’ll wait until there’s another opportunity or maybe the timing is not right because they are in the middle of some project and they don’t want to let their current employer down. That’s cool. Sometimes we’ve gone for a couple of years without actually pulling the trigger and having somebody come over, but they’ve come to Tiber training three or four times, I’ve gotten to know them really well and we’ve become pretty good friends. That’s part of it.
I’m very, very protective of our culture. I forget the old saying, but “Bad culture eats strategy and vision for breakfast,” right? That is so true. And I don’t think I’ve had a bad hire, but I’ve had one where it affected the culture a little bit. I’m the owner of the company. This is my headache, right? So if I’ve got one person that affects the culture, then that’s probably the worst problem I’m going to have and I’m very, very protective of that.

Kirill: Do you find that your existing team is also protective of the culture they’ve created?

Jim: They are, in a good way. It’s certainly not an elitist type of group. They are always willing to reach out to people that we have at Tiber training and to speak with people by phone that I’ve kind of lined up that I think would be good reviewers of some of the people that I’m talking with. But at the same time, if it doesn’t feel right, if they’re like “This person seems to be all about money” or whatever it is, then I totally respect that. When we go through, myself and our CTO and a couple of people that do speak with these people, we’ve got to all vote and be 100% yes. And if there’s one person that’s not, then there’s other people. We’ll find other people.

Kirill: Gotcha. Very interesting. I kind of wanted to move on to slowly getting into the technical side of things question. You’ve mentioned wisdom several times. That you source people with wisdom, you train them with wisdom, you approach your clients with wisdom. Could you give us an example of a problem or a project which you had recently – if you can share, of course – where you approached it and how you approached it with wisdom, how that is different to what a Big Four company would do.

Jim: Okay. Before I answer that question, I don’t want it to sound like we’re arrogant, we’re “about wisdom”. Let me just kind of define my definition of wisdom here a little bit. I get this from my wife who’s a teacher and who I really respect a lot. She’s kind of taught me that there’s three stages of learning. There’s knowledge, there’s understanding, and there’s wisdom.
Knowledge is all about intake, it’s “I’ve read this book. I keep reading the book. I read this book so many times I’ve memorized it.” But I’m a pay-per-line right? I may not even understand everything that this book says. I’ve just memorized it. So, the next phase is understanding, and that’s where I’ve taken all of that knowledge that I’ve put in my brain and I’m going to start testing it out. And I’m going to, through experience, start wrestling with these concepts. And I’m going to keep trying. I’m going to fail, I’m going to learn, and I’m going to get some understanding. And then wisdom is I’ve done this so many times now that I truly understand it, I’ve mastered this. You can even tell me things that aren’t in this book and I can give you a pretty good gut feel on what it is. I have gotten to the point where I have formed my own opinions on these things and I can clearly articulate them, I can persuade people on what’s the right thing to do. And therefore you can really help customers the best.
That’s what I mean by wisdom. It’s not that we’re smarter than anybody else or anything like that. I always want our company to be about wisdom, meaning that we have really deep hands-on over-and-over type of experience where we understand this so well. And it’s not just about reading a chapter ahead of our customer or our client and being book smart. That’s what I mean by wisdom.

Sometimes wisdom really isn’t about tech. A lot of times the technology is a little bit more about knowledge and understanding. I’ll give you an example of a company that we’ve worked with where it’s an older culture and the executives that we work with on the business side know their business incredibly well. I mean, they have been working there for 40 years. But they are not technology folks. They’re coming in from more of a “I’m going to get my PDF report e-mailed to me at the beginning of every week and at the end of the month.” But they come up to us and say, “We want an analytics dashboard or analytics application that can drive our organization from the COO down and, you know, all the direct reports to the COO and his direct reports and so on and so forth.” So how do you do that? These guys have iPhones but they have like two apps on it. The example I use is like, “Kirill, what do you want on your space shuttle?” There’s no context at all.

Kirill: I love it. That is awesome.

Jim: I mean, we just started going through and it’s like, “Okay, I can’t whiteboard this. I can’t draw pictures. Words aren’t working.” In this case, this customer is using Tableau and we just said, “Look, we’re just going to spend a couple of days just listening to you, learning your business. And it’s hard. It’s a hard business, and just trying to make sure you know we’re listening, we’re understanding, and then we’ll go back and look at your data. Data is like a sixth sense for us, we’re going to cruise through that

And then we’ll be like interior designers. Based on what I’m hearing from you and your preferences, and based on what your data says, we’re going to come up with about ten or so ‘swatches’ or visualizations that are samples.” You know, all of this is happening in a week. We’ll come back to them and show them this stuff and maybe half of them are things that they really don’t understand or don’t appreciate, but the other half, the other five they’re starting to react to. They’re like, “What’s this all about? What does this mean? That’s interesting. Can we do this for profitability instead of revenue? Can we do this for whatever?”

So the dialogue begins, and they are doing more showing and less telling, letting them react visually. So whatever they say during that hour or two-hour session, we go and change it up and bring it back to them in a couple of days. We just have this cadence of every Tuesday and Thursday we’re going to spend an hour with you, showing this stuff.

And that went on for three or four weeks until we kind of got to about 75-80%, pretty good definitions of what they wanted. Now, the backend is just Excel spreadsheets, it’s a house of cards, but we couldn’t invest time in like a star schema or ETL or any data structures until we understood what they really wanted to do and kind of began with the end in mind. And then once we’re three or four weeks in and we have a pretty good view of what that analytics solution looked like, now we can parallel lag behind, but catch up pretty quickly all of the data structures to bring it together from five different source systems and harmonize it, focus on the quality and storing the data structure that Tableau can really use very easily and elegantly. And not just Tableau, but we’ve thought about it from an entire information area, so any ad hoc reporting or any predictive models or whatever can go right on top of that and do well.

That may not be like, “Okay, we’re using wisdom to do the craziest AI type of implementation to answer questions,” but this is about changing a company’s culture to be more information analytics-centric for a thousand or more users. That’s hard. That’s a really hard problem. And that’s how we started and it’s working. You know, halfway through that process, this person that had been with this company for 40 years, I mean, that person owned this application. She had bought into it, she had owned it, she was doing the demos, and someone who is not known for their technical skills doing the demos of the application that she created, what’s more compelling than that to get people start using it? Anyway, that’s a little bit of a long story about wisdom, but that’s just an idea that we did.

Kirill: That is very clear on how you approach it, and a lot of companies would go into a situation like that and just use their most developed, enhanced tools and techniques and all their super knowledge to create some amazing powerful things which probably the client wouldn’t even understand. I’ve seen that happen lots of times where you kind of deliver and everything goes over and above the head of the client and they just kind of bin the report or the whole tool at the end of the day because it’s not what they wanted.

It’s like using a broadsword versus a scalpel, right? You go in, you have a broadsword and you like smash everything out, but that’s not what they want because they just don’t understand that and sometimes it’s better to come up with a more tailored approach. I can totally see what you mean by wisdom here. Speaking of how you’re changing the culture of a company, I completely agree that that is hard. I’ve been in a situation like that where you have to drive this new analytics focus or analytics agenda or data agenda into the company and there’s so much resistance. What has been your biggest challenge when you come into companies and you try to change the culture and you try to innovate in this way but there’s some sort of pushback? What’s been the biggest challenge for you?

Jim: I think of a couple of them. Some of them we control and some of them it’s harder to control, and they’re both huge problems. One is just the older paradigm that more reports means greater success. They strut in and say, “We’ve got a thousand reports here.” “Great! Is anybody using them?” Some of these analytics solutions—you know, I guess one powerful analytics solution should be the equivalent of a thousand reports.

And a lot of times the reports are showing lagging indicators in their monthly reports, their weekly reports. By that time it’s too late. You need to have a pathway of what are leading metrics and monitoring them on a day-to-day basis through a very dynamic, actionable, drill down from the top below, COO level all the way down where everybody can kind of see their piece of the pie, look to see what’s going on right there, take action as if the COO was right on your shoulder telling you what to do if this situation happens and then make those good decisions so at the end of the month your number is good.
That’s a very different paradigm because I think sometimes companies are like, “Well, this is only like one…” — say it’s a dashboard application that’s got seven pages to it. That’s good. A thousand people can use that and do their job and they understand it and it’s intuitive and it’s powerful and it’s dynamic and that’s good. You don’t need to have a thousand reports to prove that you’ve been successful. I think a lot of times the less is more.

And then the other thing that we see a lot of is just people’s fear of sharing their data. Well, it’s not your data. It’s the organization’s data and do you understand that if you share your data you’re going to be able to also share and receive other people’s data? And really the value is not just looking at this data silo, it’s looking at why your data silo looks the way it is because it’s probably related to some of these other data silos and what’s going on in them. If we bring it all together and we harmonize it and blend it together, you’re going to be smarter and that’s what’s best for your organization.

Some people fear that someone is going to take their data and report on it differently and make bad results come out or incorrect results. Sometimes it’s fear of job security. “You know, if I let you have my data—I spend two weeks a month putting this data together for people. What am I going to do?” You’re going to be able to analyse it now! You shouldn’t be spending 80% of the time gathering the data and 20% of your time analysing it. You should flip that around. For goodness sake, you have an MBA. You don’t want to be gathering the data.

Kirill: That’s great because I was just about to ask you why do people have this over-protectionism of their data. I’ve totally encountered that as well. People just won’t let you access their data because it’s really hard to understand why it’s happening. Sometimes, like you say, their job security depends on it, and other times they’re really afraid that you’ll screw it up. There probably is a chance that you’ll screw it up, but if everything is done smart, if there’s backups and you don’t get access to the direct data at the start but just do like copies of it to get things rolling, then it could totally be done. I think people should be more open about it. So how do you go about convincing people to give you access to their data?

Jim: It’s hard. A lot of times it comes down to our business sponsor who is able to kind of help us provide that area support. It’s a tricky conversation. It just depends. I think sometimes it’s easier when people say, “I’m afraid someone is going to take my raw data and put it through some business rules and come out with formula with the incorrect value.” It’s like, “Okay. Well, we can help solve that problem.” That’s kind of what star schema is all about. It’s taking that raw data and applying consistent business rules so we’re transforming that into information. So now we have a buffet of information where people can pick and choose what they want and the results are always the same. So you don’t have to worry about that anymore. You’re going to be the subject matter data steward on this. You’re going to be able to monitor it and you’re going to be able to define what the business rules are if they change. You’re going to be able to determine who does have access. Maybe this isn’t public or all organization wide type of information, so we’re going to respect all of that, but you do need to understand that this is the company’s information. Information is a huge asset. Your organization needs to leverage this asset and you kind of need to relinquish control so the company can use this as an asset, as it should be used.

Kirill: Yeah, gotcha. I was actually just reading somewhere yesterday or the day before that a lot of organizations acknowledge and understand that data is an asset. They understand that if all of their data was to disappear tomorrow, that would be a crisis. That would be the end of the world. But at the same time, they don’t understand that if your data is not being looked after on a daily basis or if it’s not being used to its full potential, that’s a huge lost opportunity that you’re missing out still.

Jim: Right. And even going beyond that, I mean, there’s a lot of companies that have just incredible amounts of data and they need to think how that data can be used to be monetized, to create products. I know that some of the things that we’re doing is we have some start-ups that we’re working with, and that start-up company alone has a great application, they’re collecting a bunch of data, and that may be part of the definition of the valuation of that product and that company.
But now that we have that data, let’s bolt on kind of a sister analytics product and now the valuation of that company has doubled or tripled. Those analytics become almost the report card for whether your alpha customers are using this and how successful they are to kind of prove back to your alpha customers that they are receiving the benefits that they were intended to receive. It’s also great for VCs, they’re seeing the results. Yeah, I think people need to think of information and data, how to monetize this, how to create new business offerings and products by using it.

Kirill: Yeah. I totally agree with that. I just want to move into a bit of technical stuff. You’ve mentioned star schema a couple of times, and ETL. For the benefit of our listeners, could you explain what those terms mean, please?

Jim: Yes. Star schema is a way to design data models. It has been traditionally in relational databases but it can be abstractive to really any data structure. I think our company is really trying to see how that would lay on top of Hadoop and some other types of non-traditional databases. We’re still in the infancy of that so I don’t have a lot to say. The details are there, but star schema is basically the idea that you have in the centre of it a fact table that usually is more numeric, has your metrics in it, your key performance indicators, and then surrounding this star schema are what are called ‘dimension tables’. These are the contexts of how you want to analyse your metrics.

So, anything to the right of the word ‘by’ is a dimension or context. So I want to see net profit by store, by quarter, by year, by salesperson, by product, by channel. So, all of those kinds of reference points that can either be like column headers or row headers to kind of analyse information are the dimensions. And the idea is that, when you create a dimension, those dimensions need to have a single set of values that are used by any fact table.

Usually a fact table is that the store’s metrics is coming from one source system. The idea is that I can start creating 10, 20, 30 fact tables coming from all different types of source systems. Maybe it’s inventory, it’s budget, it’s sales, it’s shipments, it’s fulfilment, it’s returns, it’s promotions. All these are fact tables that have metrics in it, but when I have a product dimension that shows the dimensions that reference to all of those different facts, I have one table. I have one list of products that are going to be applied to and connected to all of those fact tables.

And what that does is it allows me to correlate. It’s the bridge that allows me to correlate and analyse across all these different fact tables so I can do some very powerful analytics on it. That’s the idea of a star schema. It’s a way to build enterprise analytics and create kind of this buffet of information that is independent of source systems. Source systems can kind of come and go and you still have this information buffet and star schema that allows you to analyse your enterprise.
I think these things have failed a lot in the past because people say, “Okay, I’m going to spend three years and build out an enterprise data warehouse,” and you never kind of get there. These things are always evolving. By the time you’re done, after a year it’s obsolete. We don’t take that approach. We’re building out a subject area that may be one or five different fact tables with its supporting dimensions. We’re doing that every couple of months. So we always have that principle. If there’s an existing dimension, you have to leverage it and conform to it.

But what that does is like Lego blocks. We’re just kind of Lego-blocking all of these different subject areas together, releasing at least one and sometimes we have two or three teams working in parallel, releasing multiple subject areas in parallel at the same time, so just gradually taking these baby steps to create these subject areas. That’s very powerful. So that’s the idea of star schema.

And the way that you basically load this data, either the dimension tables or the fact tables, is using ETL. ETL stands for ‘Extraction, Transformation and Loading’. So you’re extracting the raw data out of the source system and then transformation can be all kinds of things. You could be cleaning up data, you could be blending the data with other source systems, you could be applying business rules and the results of those business rules, you could be adding extra columns, you could be pivoting the data in certain ways, so you can do row to row and column to column comparisons on dates or metrics or other things.
So all of that is kind of prepping it to land into a fact table or dimension table. And the fact table and the dimension table is definitely built out and designed with the end in mind about analytics. A lot of people just think, “Why do I need a second database? I’m just going to dump all the data that’s in my source system.” I kind of look at them like—that’s like saying you want me to build you a house and I’m going to go to a home depot or whatever stores you guys have in Australia where you have all the home supplies and you show me your lot and I’m just going to dump everything on your lot and say, “Good luck. Home, sweet home.”

We don’t do that, right? We intentionally think about what you want to do with this and we build the star schema intentionally with how to handle ad hoc queries, predictive models, analytic solutions, dashboards, all those types of things.

Kirill: That sounds pretty cool. Thanks for describing that. Would you say there’s situations where the star schema is not the best approach?

Jim: Yeah. I mean, when you’re developing a star schema, it takes time. You’re investing a good month or two months doing that. So, if you’re trying to just explore data, I’m definitely a proponent of having some type of data lake where you do dump the data and you put it through predictive model or Tableau or some way to explore the value of that data to see if it meets your needs. I think this is probably true with some predictive models. Sometimes predictive models, they’re just being fed from one source system that may have high quality. And if that’s all that needs to be done, great, you’re done there.
But when I started asking questions of, “Do I want to correlate this with other areas, other source data, and how frequently do I want to do that? Do I want to start supporting ad hoc queries where I need to have a very flexible, intuitive, high performance type of data model?” that’s when just dumping the data in a data lake isn’t going to cut it. So, let’s put it in a data lake, see what the value is and maybe not everything in the data lake is going in the star schema, but the ones that do need to be correlated and have ad hoc query and be very dynamic, those are candidates, as a subset, to start building out in a star schema.

Kirill: Okay. Fantastic! Thank you for sharing that technical description. I think a lot of people will find it useful. I have one question leading towards the end or wrapping up this podcast. I’d really like to know what your take is on where the field of data science and data analytics is going. As you are a business owner, your company is in the space of performing consulting and analytics services, you obviously see a lot of things that are going on in this area. Where do you think this space is going? And more specifically, what do you think our listeners should prepare for in order to be ready for the jobs of the future in this industry?

Jim: Wow! Kirill—(Laughs) If I knew the answer to that… I’ll tell you, I was actually talking to our CTO and another person on Friday about this. This is a really exciting time. There are a lot of new products coming out and have been coming out for the last several years. They’re not all going to be around. I mean, there’s going to be conformance at the end of the day in this space in all of the types of technologies that we’re using.

I’m not trying to escape here, but I think one thing that I see a lot of is that there are so many people, prospect customers, customers, people we’re recruiting, and just hype in the market that are confusing tools with solutions. At the end of the day, tools are part of the solution but people need to use these tools. They need to be practitioners. They need to have experience to develop these things and use the tools effectively.

I just look at Hadoop. I think Hadoop is incredible, powerful technology. I remember three or four years ago talking with someone pretty high up in an organization where they were like, “All right, we’re going to get rid of all of our relational databases by the end of the year and everything is going to run on Hadoop.” I’m like, “What are you talking about?” At that point, I don’t even think it does updates. (Laughs) I think we need to look at these tools and understand really not that it’s cool or that it has this feature or function and kind of be a tool looking for a solution, but to try to use tools to create solutions.
I think the other thing is that a lot of organizations really are still trying to just crawl on this stuff. They are still trying to figure out how to blend data together and have major data quality issues and conformance and just simple ad hoc queries. Old traditional BI type of products that have ran for 20 years, they still haven’t been successful just establishing those types of environments, so to come in and say, “We’re going to create artificial intelligence for your applications,” they think that’s a solution. I mean, I’ve had one company say, “We stock our stores and we want to create some predictive models on how to stock our stores so we don’t run out of inventory.” And I’m like, “Why don’t you just kind of take the average of some of these things for the last 52 Fridays and I bet you that will get you about 98% of the way there.”

Kirill: Gotcha. Okay, thank you very much. So, guys, make sure you understand the difference between tools and solutions and you know what you need to focus on in your specific case.

Thank you so much for coming on the show, Jim. I wanted to make a quick shout-out to you guys. I think you guys are building a great culture, you have a great team. I’ve spoken with Jeremy quite a lot; I’m connected with the people that have taken my courses. I think it’s great that you are encouraging your employees to diversify and broaden their skillset.
If anybody listening on this podcast is interested in learning more about your culture and possibly lives in Washington or is willing to relocate to Washington for an incredible opportunity like this, are you guys hiring at this stage? And if you are, what is the best way to contact you?

Jim: Yeah. Like I said earlier, we’re always kind of in hiring mode. We have a couple of projects that we are putting people on right now, but even if we don’t, I mean, we have a lot of inbound calls and like I said, we always like to try to get to know people and have them get to know us for several months before we pull the trigger on anything like that. We want to make sure everyone feels comfortable. So, yes, we’re always looking. The best way to contact us—I’m on LinkedIn, Jim Hadley. You can go to Tiber Solutions – tibersolutions.com – or my personal e-mail address is [email protected] You can just e-mail me directly.

Kirill: Beautiful. Thank you very much. And guys listening out there, I think this is a great opportunity for you if you’re looking for a great culture to further develop your analytic skills. And one more question for you today, Jim. What is your one favourite book that you can recommend to our listeners to help them become better data scientists?

Jim: You’re not going to be surprised by this based on all of the conversation on star schema, but a gentleman who I started to work with, Chris Adamson, wrote a book in 2010 called “Star Schema: The Complete Reference.” I am the technical editor on the book, but putting that aside, it is the best book that I’ve ever seen on star schema. It takes a very systematic approach of assuming you really know nothing and just layering in the fundamental groundwork and keep adding on layers and layers and layers to the point where it is almost like PhD level type of stuff. So that to me is the best book. In our company, whenever we have anybody come on to Tiber, I buy that book for them, I have them read it, and then when they come to me and say, “I’ve finished reading the book,” I tell them “Read it again.” So that’s how important it is to our culture.

Kirill: Okay. Guys, start reading. It’s called “Star Schema: The Complete Reference” by Chris Adamson. Thank you very much, Jim, for coming on the show and sharing all your knowledge and wisdom with us and telling us how you are innovating in the space of analytics consulting. I really appreciate it.

Jim: Thank you, Kirill. You are doing great stuff. You have just such a great audience and wide breadth of folks listening to you and you’re doing fantastic things. So, thank you for you and what you’re doing in this space as well.

Kirill: Thanks, Jim. Glad to hear it. Thanks a lot. Take care. So there you have it. I hope you enjoyed today’s podcast. We had a very nice in-depth chat with Jim about culture, about analytics, about how to build a company. And even if you’re not building a company, even if you’re just looking to join a successful organization, somewhere where you can develop your skills and enhance your career and follow your dreams, then this hopefully was an example of what you should be looking out for, an example of what companies do actually exist there and that you shouldn’t settle for less. You should look for companies like Jim’s where you can actually have the flexibility to develop any kind of analytic skills you want and become, as Jim put it, a Swiss Army knife of analytics.

And on that note, of course, if you are in Washington, D.C., or if you’re looking to relocate there, then definitely hit up Jim on LinkedIn or through his website because this is a great opportunity for you to learn. And, as Jim said, you don’t necessarily have to come and work for him. You can still just come to one of their events and learn about analytics, learn about the way they do things. I think that could be a good experience.

And also, if you are looking for consultants, then I would check out Jim’s consulting organization. It looks like they have everything sorted and it looks like they have some very good people, very talented people on-board, so if you need some analytics work, then Tiber Solutions might be a good bet for you.

You can always find all of the show notes for today’s episode at www.superdatascience.com/49. There you will see links to Jim’s LinkedIn, the company profile, and any other resources that we mentioned along with the podcast transcript. And to finish off today, I just wanted to mention how I found out about Jim and Tiber Solutions. Well, one of my students, Jeremy, he actually messaged me and told me about Jim and said that this could be a good opportunity for us to get in touch and create a great episode, and I think that’s what we have created. So if you know anybody who you think would be a great opportunity for us to connect with on the podcast, then feel free to send this recommendation to [email protected] We will gladly review them and we will get in touch with those people and hopefully bring them on the show and then you will get to hear the stories of the people that you admire. So, the e-mail, again, is [email protected] I can’t wait to hear from you. And until next time, happy analyzing.

Kirill Eremenko
Kirill Eremenko

I’m a Data Scientist and Entrepreneur. I also teach Data Science Online and host the SDS podcast where I interview some of the most inspiring Data Scientists from all around the world. I am passionate about bringing Data Science and Analytics to the world!

What are you waiting for?

EMPOWER YOUR CAREER WITH SUPERDATASCIENCE

CLAIM YOUR TRIAL MEMBERSHIP NOW
as seen on: