Jonas Christensen 2:59
John K. Thompson, welcome to Leaders of Analytics. It is fantastic to have you on the show today.
John Thompson 3:08
Jonas, been looking forward to it. I'm glad we can finally get this put together. Thank you so much for inviting me.
Jonas Christensen 3:13
Yeah, I have been looking forward to it as well. And today, we have one of my most favourite topics out there, which is we're talking about how to build excellent analytics teams. And John, you are an expert in that field. And we'll get to exactly why in a minute. Because I want to hear straight from you a little bit about you and what you do and your background. So could you please tell us a bit about yourself?
John Thompson 3:37
Absolutely. I got out of college, I was an assembler programmer. I was building all sorts of systems and as part of an IT department and I got introduced the idea of using data and information and insights as management tools. And I was really switched on by that. And I moved over to a little company called Metaphor. That was one of the first companies building data warehouses and analytical applications and business intelligence. And I've been doing the same thing for 37 years, I worked at a company called Magnify, we helped create some of the early model portability technologies and predictive modelling markup language, I ran part of IBM's consulting business, then was part of Dell's advanced analytics division. And I've just been focused on advanced analytics and data and intelligence ever since.
Jonas Christensen 4:26
Wow. So that's a very impressive background. And John, you have written the book called Building Analytics Teams. Could you tell us a bit about this book and what it's about and who it's for?
John Thompson 4:38
Sure. I wrote a book about five years ago called Analytics: How to Win with Intelligence. And the idea was that after travelling the world for three years for Dell, and meeting with all sorts of non-technical C level executives, every time I brought up the idea of analytics, you could see them tighten up a little bit, get a little bit tense. So I thought, "Oh, they don't really understand"! They're getting people from McKinsey and Deloitte and Bain showing up and telling them, you've got to spend all this money, you've got to hire all these people, and they're going to do magic. And I thought, that's just not really true. So I wrote the first book, and I got a taste of it, I thought, oh, gosh, I kind of like being an author. So three years after that, two years ago, I started writing building analytics teams. And it was just me, pouring out my ideas on how analytics teams could be started, who to hire, how to manage them, how to connect with the different parts of the organisation, and the best way to do all of it. I thought, boy, I've been doing this for,at that time, 30 years, I've made almost miss every mistake known to man, I have survived. Hopefully, I can help other people who are building and managing analytics teams to shortcut that process. So the result was Building Analytics Teams.
Jonas Christensen 5:52
So you came up with this idea for this book, and you're sort of describing going around in this environment of wanting to help executives and CEOs, but them getting bad advice from all the big consulting houses and so on, what were the typical best pieces of advice that you saw that you wanted to challenge?
John Thompson 6:09
You know, I see this happening in small, large, medium sized companies all over the world, where people are either maybe they're not told by consultants, maybe they come up with this idea on their own, or maybe they heard it or read it in a magazine or something. And it comes from all sorts of different sources. But companies seem to think that they can hire a person, just a person and say, "okay, you're going to do everything that we want you to do in data science, you're going to obtain the data, you're going to manage the budget, you're going to interface with the subject matter experts, you're going to do it all". And usually what happens is that person comes in with all sorts of excitement and promise, and they get burned out about six months into it, no one person can do all of this data science and analytics and insights, it's a team effort, and it's a corporate function, we really need to understand that. So starting off with one individual person and saying they're going to be it for an entire small business, medium sized company, or enterprise just doesn't work.
Jonas Christensen 7:11
I can relate to that. I think I was that person once and it took me about three months to run to my boss screaming for help, we did turn it into an analytics team in the end. So that was a good outcome. But I was definitely burned out after only three months in what I saw as this great opportunity, which was really just me being dumped in a hole of unrealistic expectations. So enough about me, John, because I'm interested in what you see are the hallmarks of an excellent analytics team.
John Thompson 7:41
It's a great question, Jonas, it comes in a few different forms. You know, leadership is important. Obviously, I would say that being an analytics leader, but it is crucially important that you have an analytics leader who connects well with the other executives in the company. And they talk to those executives about what analytics are, what they can do, and what it can do for the organisation. I often have conversations with C level executives, senior vice presidents, executive, vice presidents, VPS, directors, all sorts of people. And one of the things that we have to get to quickly in our conversations is that in the title of what we do is the hint, it's data science. If there is no data, there can be nothing else. We often ask people ask us to do projects for them and work for them and figure things out and predict phenomena. But if there's no data, we are not magicians, we cannot do this just by pulling things out of the air. So we need data to describe the phenomenon. That's number one. Number two, an executive leader also needs to set expectations appropriately, as you said just a moment ago, there's all sorts of unrealistic expectations about what data science can do. And we need to make sure people understand that we can do brilliant things, we can do many things, but we can't do everything. And if someone's asking us to do things that are impossible, we need to be the ones that come back and say that cannot be done. But we can do this. Or we can do this that leads to this that leads to that. So you need to have conversations with people that get everybody on the same page. So you need good leadership. That's the first part. The second part is you need a team that has diversity in all manners of diversity. I'm not talking about just gender, or age, or geographic location, you need a lot of diversity in mindset, and diversity of experience. I really like to have a team that has a set of very young people, interns, new graduates, and then people who have experience and of course, some mid-range in there as well. So I think you need good leadership. I think you need a team that has good diversity. And you need a team to understand that while a data science team is almost a magical thing, it can be it's a wonderful, wonderful organisation to be part of, it can't do everything. So you need a connection with the business with the subject matter experts, because they're the ones that understand how the real world works. And that's what we're trying to do. I mean, this is the objective, if anybody, if they only take one thing away from this podcast, this is what I want you to remember. We as data scientists are trying to model the real world as closely as we possibly can. So we can predict it, we can understand it, we can be evolutionary with it. That's our job. But if we don't have all the things that we just talked about, in a high performing analytics team, we won't get the models that approximate the real world as closely as we possibly can. That's our objective. That's what we all do. That's what everything we do is in service of.
Jonas Christensen 10:45
So you mentioned leadership and diversity there, which are, in my opinion, very important topics. So I totally agree with you. And I'd like to explore both a little bit more. And actually leadership we'll get to, because I have lots of questions coming up on that. But in terms of diversity, we hear it talked about the life and you describe some dimensions of it there. But could we dig a little bit deeper and sort of talk about what you get with diversity? And maybe what you don't get, what you lack, if you don't have diversity? What are the risks and benefits of it?
John Thompson 11:19
Sure. I'm an American. You're in Europe. I'm not sure the exact nationality. What country are you from?
Jonas Christensen 11:25
Well, here's where it gets complicated, John. Because I'm actually from Denmark, but I'm sitting in Australia. So I'm really diverse.
John Thompson 11:33
You are! Absolutely. Well, I found that geographic diversity is a great thing to have on the team. People from Denmark, think differently from people from Australia, think differently from Canadians, from Mexicans and Lithuanians, and you know, Americans. I've always found having people from different geographic and cultural backgrounds, you can say geography, you could say culture, they're proxies for each other. I think that element of diversity is hugely important. Here in America, we're very wrapped up in race. So whenever you say diversity, people often think of African Americans or Latinos. And that's, that's an important part of diversity as well. But I do think mindset, diversity is more important than the way we physically look. People that are younger, think differently than I think people that are, you know, millennials are different than I am, my wife and two children are 22 and 25. They certainly think differently than I do. And no matter how much diversity you can put in a team, you can have men, women, old young Australians, Danish, Americans, Canadians, you can't have every element of diversity on your team, you can't hire that many people, you're never going to have that many people on your team. So the way I like to augment diversity in our team is that we make sure that when we bring in consultants from the outside or subject matter experts from the inside, that they're different in some diverse way than we are. So when we're talking about a business problem, there diversity really helps us move away from the data and the math to the practical. You know, how does this actually work in the real world? Sometimes we build models, and we do something that's US centric. And then we find out from the subject matter experts that that kind of thing is limited or illegal in California, for instance, or Massachusetts or Texas. So then we have to take all that data that describes that that geographic part of the United States out of the equation. So diversity can be all sorts of different things. And people need to open their apertures when they're thinking about diversity, because it encompasses almost every element of the human experience.
Jonas Christensen 13:43
It does. And I think there's so many subtleties in it, that it actually is very hard to describe what you're going to lose out on. But you'll really see it when you start working in diverse team that you do get these inputs of "Oh, hang on, that's not how we do at home" or "No, no, that's not how you do it - that's not logical", whatever it might be. Whether you're modelling customer experience, or trying to identify certain things in images, or what have you, whatever you're using data science for, that diversity is super important.
John Thompson 14:11
Yeah. You had mentioned, what do you miss out on and I didn't address that, thank you for bringing that back to the fore. What you do miss out on is just what you said, you end up with siloed or limited or unit dimensional or maybe one or two perspectives in your thinking. And when you have baby boomers, I'm on the cusp between a Boomer and Gen X, whatever the next generation is, you know, and millennials and all these different people, like you said, people that are from a different cultural background, you bring a problem to the fore. And they said well, that we just don't do that. That's not the way we think of it, we wouldn't even start there. So that's a great way to understand that if you don't have those many minds, or many thoughts or many perspectives on the table, you're going to come up with a model or a view that just doesn't represent the world.
Jonas Christensen 14:59
And I think another subtlety that I discuss a lot with my team, which is a data science team is that one of our strengths is also a weakness, which is we tend to try and solve every problem ourselves. So we don't tend to go and ask other people for help or input because we are the problem solvers, people come to us with their problems, and we solve them for them, not the other way around. And we have to be really careful that it doesn't create this proverbial silo of thinking or ways of solving problems, where we do it by ourselves in our own time, because actually, you often don't need to look any further than maybe the guys next door on another floor or your marketing team or sales team or your HR team, they will have a very different personality style, typically, and therefore different thinking styles. Yeah, really interesting.
John Thompson 15:45
Just one more little bit on that topic, before we move on to the next area of discussion is that I was having a discussion with one of the business units in my current job. And they were saying, well, could you do this for us? And would you do this for us and those kinds of things? And I said, whoa, whoa, whoa, hold on. I said, we're doing this with you. This is your data. This is your phenomena. This is your process. We're the ones here assisting you. We're not doing this, you know, on our own, we're doing this with you. We're not doing it for you. We're doing it with you. And they said, well, when it's all done, can we have access to it? And I said, Of course, this is your data, and I'm here is someone helping you everything that we do you can have access to, but like all the data we can have access to I said absolutely. All the applications. Absolutely. You know, this is not something we're doing for you, for us to control it, we're doing it with you to give you tools to do your job better in the future. So you know, I think the whole idea of breaking down this us versus them, and it's a "we" is a very important foundational view.
Jonas Christensen 16:49
Yeah, brilliant. I like that. Now, John, I'm gonna give you a question that is very easy to ask and very hard to answer. It's a big question, and you can take it in any direction that you want. But I'm interested in hearing from you, if you were to design the perfect analytics team, what would it look like? And why?
John Thompson 17:13
That is a big question. And we can go lots of different ways, and we will. I tend to think that the analytics team, you don't need an empire, I don't believe that you need hundreds and hundreds of people to do a good job, you're gonna have a team that's as small as five or 10, you're gonna have a team that's as large as hundreds. And I've certainly had both. I think the ultimate analytics team has good leadership, as we've talked about, it has a group of people that has age diversity, at the very least. And then it has a group of people that has an understanding that each of them are specialists, each of them have an area of expertise, someone might be very good in neural networks, someone might be very good in advanced statistics, someone might be very good in text processing, or natural language processing, or natural language generation. Whatever it happens to be, or someone might be a Bayesian expert. Whatever it happens to be. I always want people on our team to possess intense curiosity, and an understanding that they can never know everything. People that come into the team that think they know everything tend to be problematic. If you come in and say, "I'm very good in this area. But I don't know these other areas very well", I love it. You know, because if you do want to learn, we will give you time and money for professional development, we will put you on projects as a second data scientist with someone who's an expert in the area. So the team needs to be large or small, don't really care about in size dimension, but it needs to have age diversity, it needs to have curiosity, and it needs to have self awareness of where people are experts and where they're hungry to learn more. I think if you have those three elements in any team, you will be very successful.
Jonas Christensen 19:04
So I'm not hearing you talk about specific role types or anything like that. Is that on purpose? or would there be certain roles that you would definitely have in a team like that?
John Thompson 19:17
Yep, that's a different way to look at it. It's a different lens. And we can certainly talk about that. I do think that you need certain roles in the organisation. I've spent a lot of time thinking about roles in a formal sense. So you need data engineers, you need people who are good at obtaining data and bringing data together, they don't really need to be that good at cleaning data. They need to be good at obtaining data and moving it but they could be good at integrating it to, you know, the roles that I always think of are can be very narrow, or they can be broader. I don't really have a problem with that as long as I have different people that can cover all the different elements of functionality that I require in my team. So data engineers are required data scientists, obviously, are required and I'll come back to that role in a second. Data visualisation experts and storytellers are very important. And then I have someone on my team now that I've never had before, who isn't an analytics person, isn't a data scientist and really doesn't understand data science, but understands the organisational element processes and the culture of the company. And that person bounces from project to project, helping the more technical people understand the softer side of projects and how to be successful. So engineers, data scientists, data storytellers, visualisations, and let's say a cultural attache, for the lack of a better term that seems to work out pretty well. Now, going back to the data scientists, we have roles in our data science, we have interns, junior data scientists, data scientists, senior data scientists, and principal data scientists. You can be in one of my organisations or the current organisation and you can go from being a college student, all the way to being an expert data scientist. And you can make as much money staying in the data science track as you can by moving over becoming a manager, probably even more. So I've made sure that data scientists don't have to give up being a data scientist just for money's sake, because they can make more as a manager, I think that you do a disservice to people if you set up your compensation structures that way. So that's the way I usually think about the roles. Does that help?
Jonas Christensen 21:31
Absolutely. I think that was very enlightening. And the salary component is really interesting, because we almost have this built in motivation or structure, or the way we try and foster talent in the organisation is that we must push them up the organisation ranks rather than becoming increasing specialists and the pay is typically tied to that. So it's actually quite inspirational. John, and I think a lot of organisations could listen to that and take inspiration from that. I have a question and that was that something that was hard to get through in the organisation that that's the way you want it to do things? Or was that sort of an easy win? Because I could see, in some organisations that would be challenging to say "I actually want to have these types of roles from a salary point of view be just as quote unquote, important as managers", and so on.
John Thompson 22:22
You know, I think it could be very difficult in some organisations. I was fortunate in this last organisation that they were going through an exercise of completely re-architecting job families at the same time. So when I came in and started talking about all the different roles that I wanted to have, and the differentiated roles, and the salary bands and the two tracks, technical and management, they were very open to it, they saw it as an intriguing way to look at technical talent. They hadn't done it before, I had to do a lot of talking and explaining. I didn't have to do much selling. I just had to really bring everybody along so they understand what I was trying to do. There were a couple people that were sceptical, and said, you know, these principal data scientists are getting paid a lot of money. And I said, Yes, they are. And they will. And they do. I said, but one application that we build returns 10s of millions of dollars to the business, and they said, "Really"? And I gave them a couple examples. And that was the last conversation I had on justifying data salaries for data scientists.
Jonas Christensen 23:27
Yeah, brilliant. I remember one of my data science colleagues many years ago, saving an organisation $20 million, and they got a $200 gift voucher for the local hardware shop, so you're definitely doing better.
John Thompson 23:41
I mean, it's laughable. It's almost sad and embarrassing, but we can laugh about it here in this situation. I've seen it over and over again, the people at the top the C level executives, are I getting bonuses, I get bonuses. I'm not saying I'm not well paid, I'm very well paid. But these people are doing things that are saving millions of dollars, and they're getting an attaboy in a corporate recognition system, which is totally out of whack. I've never even tried to attack that maybe sometime in the future. I will.
Jonas Christensen 24:08
Yeah, small wins first, and then we can grow from our successes. John, the other one was this, you call it a cultural attache, or I think sometimes we hear the term, an analytics translator. And it's a role that you see pop up here and there. And in some cases very deliberate, other times it's maybe more experimental. I'm interested in hearing from you how you realised that you needed this role, because it sounds like you had lots of years of not having this and this is actually a new thing. So you must have had some sort of aha moment where you thought, "Okay, this is the role that we need". And then after that, now that you've got it, what is it given you that you're really excited about?
John Thompson 24:49
It's a great question. I'm going to take it in just a slightly different direction, if you don't mind, Jonas. I think it was McKinsey, or maybe even Davenport, or them together, they came up with this notion of an analytics translator. I grew up in the software world after my first job as a an IT developer, a role in large a corporation, I moved over and have spent most of my career in software companies. And I think of myself as a product manager, I was growing up as a professional and trained as a product manager. So product manager, one of their most important functions is translation. So translating between the technical side, the developers and the business side, people giving you the requirements and coming together with product and product plans, that's been my life. So when they talked about an analytics translator, it resonated with me very deeply. And what I do with my teams, my data science teams, and this is something that people freak out a little bit about when they join one of my teams, is that every data scientist needs to be an analytics translator. And they're like, "whoa, hold on, hold on. I'm an expert in neural nets, or I'm good in classification, or clustering" or something like that. And I'm like, "Yes, that's part of your job. The other part of your job is being a communications expert, you know, talking to an executive talking to a subject matter expert, getting the requirements, building the systems, and then explaining to the people who are going to own it, what they're going to get and how they're going to operate it". So every one of my data scientists, and my data engineers are analytics translators too. Now this cultural attache is something completely different. I mean, the person on my team is from Europe, he speaks many different languages, we just had a project where there was a requirement to understand source documents in German. It happens to be that this guy speaks German. So we couldn't have built the model that we're building now without him. Because all the source data is described in German, none of us as Americans knew how to read it. He went in, read it, explained it to us, and we're like, oh, this is perfect. This is exactly what we need. So an analytics translator is something I expect, as a sub function of every data scientist. Now, this cultural liaison is just something completely new and different and unusual, that I never even thought of before. You know, this situation landed in my lap, the person on my team is a wonderful individual, I really enjoy working with him. But he brought things to the team that I didn't even know we needed.
Jonas Christensen 27:21
Okay, John. So one thing is the types of skills you need an analytics team. What about personality styles? Is that something we should consider in our hiring and promotion process?
John Thompson 27:34
You know, we touched on this earlier, and we'll go a little bit deeper here. In my team, what we're looking for and what I've imbued in everybody that's in the hiring process. And everybody on my team thats involved in hiring processes that we're looking for people who are curious, we already stated that we're looking for people who are honest, we're looking for people who are ethical, we're looking for people who are not afraid to bring up difficult conversations. Sometimes people don't, for the most part, people outside analytics don't understand that advanced analytics is more of a creative endeavour. It's not like an IT function. It's not like developing the next data input form, it doesn't work that way. You know, we often are looking at a wide range of data sources and trying to bring them together in interesting ways, as I said, in service of trying to model the real world, in a computing environment. And that is not a straightforward linear process, there will be failures, I push my analytics teams to experience failure. And if you're not comfortable with yourself, and you're not self confident, I'm not saying arrogant, but just comfortably self confident, you will have a hard time coming to the team and explaining that failure, because many people been raised who are in analytics that they never fail, that they're brilliant. They're the smartest people in the room, and we're in the world even. And failure is not possible. Failure is not an option. Well, I can I'm here to tell you, if you're doing things right on my team, you will fail many different times. So we're looking for people who can come to the table and say, I've been trying this for two weeks, and I can't get it done. And having the team help them say hey, did you look at this kind of data? Did you bring this in? Did you try this technique? Did you integrate the data this way? How about your feature engineering? You need to be humble enough to listen to others and have them help you. So those are the characteristics. We're looking for curiosity, ethics, honesty, integrity, diligence, and self awareness to be able to bring failure to the table.
Jonas Christensen 29:39
Nice. How do you test for that when you look for people?
John Thompson 29:42
you know, we try not to be overly intrusive in our hiring process. You know, we try not to have some companies have so many interviews, it's it's ridiculous 10-12 interviews talking to all these different people. We usually have a screening interview, HR and talent acquisition helps us with the screening. And then if they get past that, then they have an interview with one of the team members, one of the data scientists, if they get past that, they get a battery of tests to take home and do and you can do them however you want, you can talk to your friends, your daughter, your dog, your wife, your colleagues, whoever can help you get the test done. Because in the real world, you're gonna have all those resources to so if they pass those tests, and they have an interview with a panel of people, and if they get to that panel interview, then they meet me. And if they get to pass me, then they get an offer.
Jonas Christensen 30:28
Yeah, good process. I like doing the test myself too, as a take home test. Because I agree with you. It's it's a real world scenario where you do have the internet there to Google things. And you do have friends and family to ask and so on. The test where you get sort of prompted on the spot and that, there can be all sorts of things happening on the day that can throw you off, and that you're not actually performing as well as you would have otherwise.
John Thompson 30:51
Yeah, maybe maybe you had a bad sandwich, or you know, maybe you had an argument with your wife or your daughter, maybe your dog bit you, and I can't control what's happening. And if I put someone on the spot and say, you know, in this next seven minutes, I'm going to determine if you're worthy of my team, I find that unrealistic.
Jonas Christensen 31:09
Yeah, and this whole process is so full of bias. I've actually got a whole episode on that from a few weeks back, where we discussed just how full of bias our hiring processes are, because we are humans, and we are biassed at the core. That's how it is.
John Thompson 31:25
Yeah, I personally have taken a stand that I have a job and I'm happy in my job, I'm not interviewing or anything like that. But when I am interviewing, if they asked me to take IQ tests or behavioural tests or anything like that, I just end the interview process. And they say why, you know, you can clearly pass these tests, you can be on the team. And I said, this shows me that you want to have a team that has cognitive cloning, that everybody thinks the same. Everybody can pass the same test, everybody's the same person. I don't want to work in an organisation where everybody thinks the same way.
Jonas Christensen 32:00
Yeah, it's very interesting, such an interesting topic. Because we as analytics, people will want to put metrics on things and want to put numbers on things. But sometimes it's actually not possible. And we do need to foster those interpersonal skills to kind of pick up people. Now another thing you said that made me reflect was this idea of experimentation, failure and curiosity, because you're talking about it within the analytics team. But there's actually also, I suppose, and outwards relationship between the analytics team and the organisation where this is a bit of a challenge, because the way I see it for a lot of people outside analytics or data science, those projects that we do, they sort of look feel smell a little bit like IT projects, and it projects typically have a defined outcome. And you put this piece of software, and this is what you're going to get, and they'll take so long. Now, of course, we always go over budget and time, that's another story. But at least there is an outcome. And software in historical context also is you put a and b in and you get c out, always. We're different, we're experimenting, we don't know what we're gonna get, we might not get anything. And when we build pseudo software solutions, which are algorithms, we also might not know what we get, how do we solve that and sell that into the organisation? It's a long question with a short ending.
John Thompson 33:25
I always think that the longer the question is shorter the answer, so the answer is five. No, I'm kidding. No, it's a great question. And it's something that we as analytics professionals really need to internalise and understand that this is a challenge for us. And that I often start out in a baby step kind of way. First thing I ask people is, do you just want a number? Do you just want a number, if you do, then we will get you a number. And that'll take us two weeks, three weeks, a month, whatever it is, and we'll give you the number and you can come back to us if you have any more curiosity. Or if you want to change the way you do business and you want to continually improve. The first one is a project we'll do a project for you. If you want to continually improve, that's a programme. And that's something we'll be involved in from now until some point undetermined point in the future. And that that will take a lot more time on our part and will take a lot more time on your part as well. Now, tell me, do you want a project or programme? And often people say I just want to project like, okay, great, we'll get you that number. So we'll work with the subject matter experts. And we do that kind of stuff. Now, if it's a programme, and it's something that and even in a project, I suppose you were alluding to earlier, the creative nature of modelling and we talk to people about that. And we say, "look, we may not come up with anything, we may get into this and come up with nothing". But the first step that we always do is we always do an exploratory data analysis. So we take the data that we know and we bring it all together and we integrate it and we look at the norms and the distributions and the statistics and the averages and things like that. We sit down and we'd say to the people, the subject matter experts, the business people, "this is what your business looks like. This is what your data describes to us". That alone often is a game changing moment for people. Because many people have been working on a business for 5-10, seven, 8-20 years, whatever it is, and they've built up an intuitive sense of what the business is, now the business may have moved on, it may have evolved. That alone, right there, is a very important thing to do with your subject matter experts, and will often give you and your data science team a level of credibility that you didn't know you didn't have. But we'll have forever forward with that person. So after that, then you get into it, you do some more analysis, you do some more exploration, and then you come back to them and say, we can build a model that has a high degree of confidence that will help you improve your business. That's the great outcome. And then they say, "Okay, how soon can you do it"? I usually give a fudge factor of about three months in any project. And they say the same thing that you just said, Jonas. They're like, "well, any other IT project, they tell us, it's going to be done on this day for this dollar at this time". And I'm like, "we're not an IT project. We're an analytics project, we have many factors, we have data and computers and all this other kind of stuff, but we're different. And sometimes we're going to be successful, sometimes you're gonna fail, sometimes you're gonna have to take different paths". And often what I do is I tie the outcome to $1 amount, Australian dollar, US dollar, euro, whatever currency you want it to be, you know, we're doing a project right now that when it's done will return somewhere between 24 and $35 million a year to the company. Now, when I say that, and I say it's gonna be between January and March, they're like, I don't care. If it's done in July, I don't care. So there's ways to bring people along in incrementally, and there's a way to align your interest with them. But they don't really care. I mean, most of the time, they don't care if it's done in January or March, they just want to know it's gonna get done. And if you let them know, upfront, I can't tell you it's gonna be March 15. It might be April 15. But for $34,000,000, 30 days is fine.
Jonas Christensen 37:18
So listeners you can hear, John is saying it's as much about the communication as it is delivering the actual result, very important. Now, there was a lot more in that comment, by the way, John, there's a lot there a lot there. Now, John, I've saved the best for last, because this show is called Leaders of Analytics. So we're going to be talking about leaders of analytics now. And in your book, you describe that proficient analytics leaders typically possess the following traits and I'm going to read your list. They're optimistic, yet sceptical, intensely curious, mostly introverted, logical, a combination of left and right brain orientation, at the same time, intelligent, self critical, prone to perfection, social but reserved, and in some cases, they appear to exhibit a lack of focus, or possibly too much focus. So other than describing me, what is so unique about managing analytics teams? And why are these the common traits of really adept analytics leaders?
John Thompson 38:58
These are the traits that I've observed in people like you and me, over the past almost four decades, these are the people that have been successful, and have had longevity in analytics. When you look at the people that have been around, this is a combination of traits that they all exhibit. It's really interesting. You know, I did an initiative a couple years ago on neuro diversity. And I think we're going to find that, you know, the people who are normal are in the minority. It's one of these things that especially in analytics, we see a lot of neuro diversity, people that are on the different spectrums from all different kinds of disorders that people describe. But, you know, I think that those are just elements of focus. That last I think it was the last one that you read, either people exhibit too much focus or not enough focus, and that's usually a DD or ADHD or something like that. And, you know, it's it's one of those things that we have a very rich community here, and a wonderful group of people that have all sorts of interesting quirks and differences. And, you know, there's one story I tell about someone that I was actually given to fire. And I spent some time with this person, I got to know them very well. And it was just the environment they are in didn't work for them. We changed the environment and their quality of the work not only maintained, but went up dramatically. So you know, those traits, you know, it's a guidepost. So if you see someone like that, they're probably a kindred spirit. You know, it's probably someone that you can embrace and work with closely. So it's just something to keep an eye on.
Jonas Christensen 40:31
Yeah, I said, Ha, when I read that list of what it was, it was funny, because I could relate to it so much myself. But what do you think is the thing that attracts this type of people to this type of work?
John Thompson 40:45
You know, it's something that was said, I think one thing that's very, very important and almost universal for all of us is that, and I went to college in the United States. Early on in the computer science movement, I went to one of the first schools in the United States that had a computer science department. And I got a four year degree in computer science, which was unheard of at the time. And years later, I went back to that university, and I'm still working with them to help them improve their analytics programme and imbue communications and softer skills into the curriculum. And I asked them, I said, "What are we looking for in kids"? And I was a kid at the time, and kids that came to the programme, and they said, "We looked for kids that were tinkerers, kids that worked in factories, or worked in automotive shops or something like that". I said, "why"? And they said, "well, because in computer science, you're always tinkering". You know, I bring it back to curiosity is what I do. Curiosity, and focus and diligence. You know, those are the traits that bind all of us together, we want to tinker, we want to play, we want to tune in and mess with things until we get it right. If you don't have that as one of your core elements of your behaviour, you will not be successful in computer science, you will not be successful in analytics, you could carve out a career and you'd be okay. But you won't be great. So people that are curious, and then want to tinker and have the fortitude to stick with the hard problem. Burning brain cells, as I say, those are the people you want around.
Jonas Christensen 42:14
Yeah, interesting. So if I combined the tinkering with this sort of appearance of either lack of focus or too much focus, I can see that sometimes I can imagine situations where that's at odds with where the organisation wants to go. So how do these leaders sort of set themselves up to meet the expectations of a large group of business stakeholders, which it typically is which are often impatient and, you know, these sorts of impatient execs and, and senior leaders that want, they want that number wow, "we want the outcome now, tell us, our sales are tanking, we want to know now" all that sort of stuff. What did they need to do to work well, in that environment?
John Thompson 42:54
You know, it's almost counterintuitive. You as an analytics leader, I as an analytics leader, need to be the buffer to the team, these demanding executives, these people that are impatient, they're all good people. They're tasked with doing things, and they must get their work done and hit their objectives. But it's our job to keep their impatience and their demanding nature away from our data scientists, those people can come and talk to me every day, and I'm happy to talk to them. But they don't get to talk to my data scientists, the subject matter experts do but these executives do not. And would the counterintuitive part of it. That's the protective nature of the answer there. The counterintuitive nature is that you do more rather than less. So each data scientist I write about it in my book about their personal project portfolio. Each of them have two or three major projects, probably two major projects two or three smaller projects, and they do service request too and what happens is that you have the data scientists producing constant streams of results. So there's always someone in the organisation who's going to be unhappy. They didn't get what they wanted, or the answer didn't line up with their perspective, or whatever it is. But there's also a corresponding number of people who are elated with the results they got. So you're always managing disappointment and despair with elation, and it's a static nature. So it's one of those things that you and your team do more. And you balance out, in average out the sentiment on how people view you.
Jonas Christensen 44:18
Yeah, I really liked that answer. And especially the leaders role in in being that buffer I think a lot about for my team, the I'm sort of making it very abstract here. But the concept of managers time versus makers time, you might have come across that concept before, which is someone who's on a manager's schedule, they will have their day chopped up in bits, and they can't really have focus on deep time, deep focus time to sort of develop things. But analytics teams they typically need that peace and quiet to actually get under the hood and figure things out. And what we're talking about here is not necessarily execs putting meetings in the diary, but it can be just as distracting to have a plethora of different items on your to do list from lots of people that sort of fly around in your head. So it's that buffer is super important.
John Thompson 45:06
Yeah, and it's very interesting. Jonas, I've had a number of people say to me, "I want to talk to the person who's building this model". And I will often broker a meeting for them, and I will be there. And then after the meeting, more often than not, not every time but more often than not, that person will come back to me and say, "that was a waste of time, I understood what they were talking about to some detail. And I know that they're working on it, and I know what I will, I will get what I want. But I won't ask you to do that again". And I'm like, "Thank you"! You know, because talking to someone about the inner workings of 1000 level neural network isn't very fun.
Jonas Christensen 45:44
No, you wish you'd never asked.
John Thompson 45:46
Yeah. And sometimes they ask so much. I'm like, Okay, you can have that conversation. And then after they have it, they're like, I never want to do that again. I'm like, okay, then don't ask me again.
Jonas Christensen 45:55
That's funny. That's good. Well, there you go. Problem solved, potentially. Now, John, we're almost at the end. So I have three questions left for you. One I'm really intrigued to hear your answer to is one about your new book, because anyone who has been following you on social media over the last few months, well, now that you've been very busy writing this book, could you tell us a bit about this book and when we should expect it out in the market?
John Thompson 46:23
Yeah, thanks for the question. Jonas, really appreciate that. Yeah, the book is working title right now is The Future of Data. And it's all about, it's of interest to anyone who goes on the internet. And I'm not sure I know anybody who doesn't go on the internat, I'm sure there's a few people, but I'd probably don't know them. And it goes and talks about the history of the data environment we live in now why we have the data environment we have who owns the data, and why people believe they don't own their own data, then it goes into you know where we are today, with companies like Facebook and Netflix and Amazon. And then it goes into what's going to happen in the future and how each of us as individuals can own protect and monetize our own data in the future, and what your data dividend will look like, you will be paid for your data, you will be paid for the use of your data, and you will be able to set that price of each piece of data that you own. And that's what the book talks about and explains
Jonas Christensen 47:14
Fascinating. I can't wait to read that because there are so many ethical conundrums around data and privacy that we are very far from solving. We haven't really sort of grasped as as humans yet. And then most people ignore it to be video. So that'll be a fascinating read. Thanks for writing that on behalf of humanity. John, I really appreciate that. Now, second, last question, John, I'm going to ask you to pay it forward, because we want to hear who you would like to see as the next guest and leaders of analytics, and why?
John Thompson 47:49
Have you had Schmarzo? Yeah, Bill Schmarzo.
Jonas Christensen 47:53
I have not had Bill on here. What a good idea.
John Thompson 47:56
Yeah, I love Bill. Bill's great. He and I have known each other for nearly 40 years. He has some wonderful ideas on the economics of data and how data as an asset plays a role in corporate life and all of our lives. So I would suggest bill, and I always enjoyed listening to Bill.
Jonas Christensen 48:13
Yeah, brilliant. I will read that reach out to bill in due course. Thank you for that recommendation, John. Now lastly, where can people find out more about you and get a hold of your content?
John Thompson 48:24
LinkedIn is where I share everything. John K Thompson and analytics, just look for me there. If you are in the data and analytics field, please feel free to request a connection. If you're in the field that connects with everyone, if you're trying to sell me a franchising system, I don't connect with any of those people. But LinkedIn is the best place to find me.
Jonas Christensen 48:44
Yes, we are getting overwhelmed with all the software salesman at the moment, it's crazy. I don't know how to answer them all. Well, I know what I'm doing. I'm not answering. That's the problem. Some of them are nice, but I can't answer. Too many. Okay, John, this has been a wonderful interview. And we could have gone for much longer. There's so much to cover in this space. We have to do it again. Maybe we'll have to do it again. When your new book is out. We'll maybe do a follow up and we can hear all about that book. But until then, thank you so much for being on here today and all the best with your new book. I know it has a little bit of typesetting and things to do before it's published. We can't wait to see it on the market and thank you for contributing to the global analytics community the way you do.
John Thompson 49:27
Thanks, Jonas. Really appreciate the invitation.
