An Overview of Successful and High-Performing Analytics Teams
Say something once, why say it again?
Given that we are just starting our journey together to explore the topic of this book—the building, managing, and ongoing success of high-performing advanced analytics teams—let's ensure that we are on the same page. Let's take a moment to set the stage, to synchronize our understanding of what we will be examining and discussing in this book.
My primary motivations in writing this book are to:
- Serve fledgling and experienced front line managers in the field of advanced analytics by helping them avoid mistakes of the past and to assist them in taking the appropriate paths to building sustainable and synergistic teams that can and will engage productively in the exciting process of building these new levels of machine intelligence, and
- Help senior managers and executives understand the investments needed, the timelines required, the problems that can be addressed, and the value to be derived from including talented teams and appropriate managers and management in their organizations.
Throughout this book, we will be talking about teams involved in building advanced analytics and artificial intelligence systems. These are systems that learn and improve over time. We are not talking about static business intelligence applications, dashboards, or reports that outline, visualize, or describe the past. No matter how often a dashboard is refreshed, even a real-time interactive dashboard is still a representation of the past.
Throughout our dialog and this book, the terms advanced analytics and artificial intelligence will be used interchangeably. For clarity, advanced analytics is a broader term encompassing the practice and use of statistics, machine learning, simulation, and optimization. Artificial intelligence refers to machine learning and other analytical approaches that learn from data over time.
In this book, our discussion of advanced analytics systems will encompass analytical applications, individual models, ensembles of models, systems, platforms, and cloud, on-premises, and hybrid environments.
Our discussion will also outline analytical applications, models, and environments that are built on and utilize the following analytical techniques: descriptive statistics, Bayesian approaches, mathematical principles and theories, artificial intelligence (AI), machine learning (ML), simulation, and optimization. Our discussion of advanced analytics will be as broad as possible.
Don't worry—this is a non-technical treatment of the topics. We will not delve into the finer points of ML or any of the subject areas listed here. If you are interested in the technical details of any of these fields, there are numerous consultants, experts, pundits, academic papers, presentations, conferences, symposia, books, and classes that you can engage with to enrich and deepen your technical knowledge. There are too many organizations and events to list or refer you to, but for a solid overview, you might want to start your line of inquiry by using online learning platforms like Coursera, Udacity, or Udemy. Given that those platforms are aggregators from some of the premier universities from across the United States and around the world, you will find much of what you seek from a technical perspective on those platforms.
A few words about what we will not be discussing or what advanced analytics and AI is not. We will not be exploring the topic of sentient machines or artificial general intelligence. These areas of development and topics are interesting to me, and many people, but they are more in the realm of science fiction at this point. We will be focused on the topics of building an analytics function in your organization and how to build and manage a high-performance team. Advanced analytics applications are focused on describing, predicting, prescribing, simulating, or optimizing the immediate present or the future. Artificial intelligence systems are dynamic; they are like a rocket in that they are always off course, but continually course correcting. The teams that build these systems know this.
They know that they are building living and live systems. They know that they must consider a staggeringly wide range of scenarios. They know that the systems that they build can dramatically improve business operations and, in the end, the results of those operations.
Let's start to examine in detail the factors and forces that are affecting advanced analytics in the general market, looking at jobs created, technological evolution, level of success achieved, public perception of the value of advanced analytics, government regulations, and more.
The future of jobs and AI
The question I am asked in almost every internal and external presentation that is attended by more than a handful of people is something like, "How many and what type of jobs will be eliminated by AI?"
The first few times I was asked this question, I brushed it off with a brief answer, assuming that it was a passing curiosity by the person asking the question and there was no real concern or emotion behind the inquiry. I was wrong in that regard.
This question is on the minds of many people and it is weighing on people as a real concern. In the past year, I have been asked the same or similar questions in presentations and discussions in Australia, United Kingdom, Germany, Switzerland, and the United States.
Rather than continuing to brush aside the question, I have started answering the question with one of the many studies that has proven, again and again, that AI and related technologies and systems are net job creators in the short and long term. Let's examine a few of those recent studies.
AI is an engine of job creation
One of the relevant studies is from the World Economic Forum's Center for the New Economy and Society, The Future of Jobs Report, 2018. The report includes research and findings that illuminate and explain the detailed job changes that are expected to be seen, country by country, on a global basis. The report suggests that the new jobs created will be significantly larger in number than those eliminated, and those new jobs will be higher paying and have a more secure future.
A 2018 report from the World Economic Forum (WEF) even suggested that, while we may displace 75 million jobs globally by 2022, we'll create a net positive of 133 million new ones. The WEF believes — with the current data in mind — that robots and algorithms will improve the productivity of existing jobs and create several new ones in the future. Perhaps future workers won't get a job — they'll create their own. No amount of angry hand waving or puerile legislation can stop this. We cannot even begin to fathom some of the otherworldly technologies and new career fields that'll one day arise. 
I have experienced this exact dynamic in the workplace. In one instance, people were hired to execute a rather dull and rote process in the finance department to move data from one system to another. The people hired were young, smart, eager, and willing to learn. They discovered that the organization had licensed robotic process automation (RPA) software.
The young staff members took it upon themselves to learn the software and become proficient in automating the repetitive processes, thereby eliminating the jobs that they were hired to do. Did they lose their jobs? Yes. Did the company recognize their initiative and talent? Yes. What do they do now? They automate manual processes across the company.
Now, the company has fewer openings, and no-entry level staff members to execute manual data movement, but the firm now has a number of open positions around the world for entry-level staff members to build automated data movement processes in RPA software. Previously, the data entry roles were lower paying, dead end jobs with few to no development paths or planned ways to move up in the organization. Now, the jobs are entry-level analyst roles with higher pay and a planned path to a better job and an explicit development plan.
I am aware that the previous example is not an AI case, but many organizations cannot grasp the leap to AI without taking an easier first step in an area like process automation.
As a manager or executive who is interested in and wants to drive change, you must be aware of the ability of your organization to understand, enact, fund, and assimilate change. You may want the organization to begin operating like a top tier firm in relation to advanced analytics and AI, but the organization may be run by fast followers, laggards, or even worse, luddites.
Keep in mind and look closely at the people who are the senior executives in the firm; how did they get to their positions and how long have they been in the firm? More than likely, they will be the gating factor in how quickly the organization changes and how the organization changes.
I am betting that once you take a close look at these people, you may want to recalibrate your ambitions regarding the timeline to achieve success with AI and related technologies.
Many jobs will never be changed by AI
Numerous people ask me if there are any jobs that will not be automated out of existence by AI. Rather than asking this question, I think it is more insightful and helpful to ask, "Why are there so many jobs today that have not been automated away?"
The essence of the problem can be found in Polanyi's Paradox. Michael Polanyi, a British-Hungarian philosopher, stated, given that "We can know more than we can tell, we shouldn't assume that technology can replicate the function of human knowledge itself." 
We humans operate on and with a substantial amount of tacit knowledge that we have a very difficult time expressing to other people. One of the core elements of automating a task or replacing a person with AI is that we need to understand and describe what the job entails at a sufficiently detailed level, in order to replicate the job with automation tools and/or with AI. Without this ability, we cannot automate the task and we certainly cannot expect AI to undertake the work. One timely example can be summed up as, just because a computer can know everything there is to know about a car, doesn't mean it can drive it.
In late 2019, Rob May posited a related idea. In reality, advanced analytics will create whole new industries, or at least subsegments of industries, where people who can afford the services will seek out human curated goods and services that have a high degree of creativity and customization. These services and goods will be sought after because they contain an element of elegance or personalization that is only possible through the involvement of human thought, expression, and craftsmanship. 
Regarding the jobs created in the aforementioned category, there will not be a significant number of jobs that will move the employment numbers in any one country, and there is no hard data to back up this claim, but I do believe that May is correct in his core assertion. AI will not create deeply personal experiences.
AI will predict outcomes and it will make operations more effective and efficient, but it will not deepen most, if any, experiences for people. The lesson to learn from this example is that there are a number of market segments that will be created for industrious people. With these, they will serve firms and individuals in ways that are made more valuable by being in opposition to the mass change created by AI.
These types of jobs and businesses will be small, but the prestige and expense of engaging with these firms will be extremely high. These firms and offerings will be the opposite of Amazon, Walmart, and other firms that operate high-velocity, low-margin businesses. The offerings from these companies will be deeply personal, highly connected, coveted, limited, and very expensive.
If AI will create more jobs, let's prepare for those jobs
In my view, the bottom line on employment is that AI will create and enrich jobs in a net positive manner on all accounts. In some cases, maybe not the job you have today but there will be lots more jobs available because of AI.
AI will drive change in the job market. Foundationally, the changes will be that the rote and robotic elements of jobs will be automated away, and those elements will be accomplished through software and hardware. A relevant and salient aspect of automation and the enablement of systems through advanced analytics is that humans and machines are good at different sets of tasks. Jobs will lose the robotic and mechanistic elements of work and they will gain, or become more focused on, the elements that people are good at, tasks like creative thinking, writing, presenting, and collaborating.
AI Will Create 2.3 Million Jobs in 2020, While Eliminating 1.8 Million. The number of jobs affected by AI will vary by industry; through 2019, healthcare, the public sector, and education will see continuously growing job demand while manufacturing will be hit the hardest.
Starting in 2020, AI-related job creation will cross into positive territory, reaching two million net-new jobs in 2025. In 2021, AI augmentation will generate $2.9 trillion in business value and recover 6.2 billion hours of worker productivity. 
Jobs will evolve and AI will drive that evolution
My experience with job evolution is a personal one. I started out working for my father as an automotive mechanic. I moved on to working on farms and eventually, I ended up running mills, drill presses, and lathes in factories that fed the automotive and defense industries in rural Michigan. As a teen, I saw that jobs were being eliminated and families were in distress as the old manufacturing base in Michigan was contracting. It is unlikely if you are reading this book that you or your children are working in these types of jobs, but the changing dynamic of work that we are discussing in this book is the same dynamic that I was faced with when I was 18 years old.
We need to be aware that when economists conclude there is no evidence of overall job losses that can be directly attributed to any one technology or market evolution, including AI, they are talking at a global, macroeconomic level. Of course, the effects of AI and automation will differ from region to region and from country to country.
This has been mostly as a result of the use of robotics and automation in the manufacturing sector, which has displaced large numbers of workers. In some cases, those former manufacturing workers have found employment in the service sector.
But there are also pockets of left-behind communities in parts of the developed world, where several generations of families are adjusting to life without work. 
At the macro level, there will be more jobs and better jobs for workers and employees, but let's not gloss over the fact that real change will occur, and is underway. A recent study found that a growing proportion of manufacturing jobs now require a college degree:
More than 40% of manufacturing workers have a college degree today, according to a Wall Street Journal analysis of workforce data. That's up from 22% in 1991. If growth continues at the same pace, college-educated manufacturing workers will overtake the number of workers with a high school degree or less within the next few years, the Journal found. 
People around the world need to be aware that change is constant, and it has been this way for generations. It is hard to look forward and determine the jobs that will be in demand in 20 to 40 years. Yes, I did mean 20 to 40 years. If you are reading this book, it is highly probable that you will not be working in 20 to 40 years, but your children and perhaps their children and the staff members that you are mentoring will be working. You want to be able to help them, to counsel them on where the future is heading, and this will be in relation to employment and a fulfilling and engaged career.
Looking decades out into the future is challenging for some and invigorating for others. I vividly remember telling my mother that I was quitting my solid and reliable job as an automotive mechanic at a local Ford dealership to attend college, to study a new field called computer science. Her response was, "You are ruining your life." It was hard for her to look out into the future.
Luckily for me, I ignored her advice and counsel. Now, my son, who recently graduated with a degree in computer science, and our daughter, who is studying data science and user experience design, joke, "Those computers, I think that there is a future in working with them." Each time we say it, we laugh and smile knowingly.
AI and data serve us, not the other way around
Another dynamic to keep in mind, at least in the United States, is that the social safety net is not in place to take care of everyone's economic needs in full. For the most part, people in the United States will have to work longer and, in many cases, they will not have the traditional or historical retirement at the end of a fixed number of working years. Finding work that is intellectually stimulating, engaging, and that will not be automated away or redesigned to eliminate human involvement will continue to present an enormous challenge for everyone.
Remember the old saying, find something that you love to do, and you will never work a day in your life. Hackneyed, but there is a kernel of truth in that statement. If you enjoy or love what you do, you will engage in it almost effortlessly, it will not tire you out, you will do it for longer, and you will have the zeal to evolve the role to fit the needs of the market today and in the future. When I think of my work and profession through this lens, I can see being involved in the market and industry for another 20 to 40 years, and still being passionate about my work and contribution.
I worked at IBM in the 1990s. My manager was Dave Carlquist. Dave was, and is, smart and driven, and possesses substantial emotional intelligence, far more than I have or will ever have. Years later, I was waiting to board a flight in Chicago's O'Hare airport. We were all standing there looking at our phones, and I looked up, and Dave was standing next to me. I smiled and poked him with my elbow.
Dave looked up and we laughed and started to catch up on the events of the intervening ~15 years. We were about to board the plane and Dave said to me, "You know, when you were going on and on about how data and analytics were going to be the lifeblood of all organizations, I thought that you were out of your mind." I smiled and said, "Not out of my mind, just early to the party."
These societal, economic, and technological factors and trends point out that the following are useful premises to keep in mind:
- Human creativity will not be taken over by AI; in fact, just the opposite will be true. Human creativity will be more valued and valuable in the future.
- Collaboration between people cannot be automated away. Again, just the opposite will be true. Technology will facilitate better collaboration, but the essence of effective collaboration will become more valued and valuable.
- Communication skills in all forms will become crucial and more valuable.
- User experience (UX) design and construction, the interface between technology and people, will become more important. The UX will become paramount in the engagement of people with systems, applications, and platforms.
- Advanced analytics and AI developers will be in high demand.
- AI will make simple, transactional interactions more efficient, but will do little to enrich sophisticated, nuanced interactions.
- The technology roadblocks that we face today, and some of the very recent past, like the efficiency of machine learning models to be effectively trained, computing capacity, natural language processing, quantum computing, and a wide array of other issues, will be solved in the near future.
Change is constant – aim as far as you can see
The farm job that I had when I was 16 doesn't exist anymore, but there are farmers and farm workers who are working each day to bring food to the market. Farming is a widely varied industry. From artisanal, organic family farms to large corporate organizations running farming operations, farming still exists, but for some, it is not the farming that their grandparents were engaged in, and for some people, that is a loss, but not an inevitable outcome.
Find what you are passionate about, look far into the future, find the intersection of the two, or the multiple relevant intersections of trends and evolving markets that you care about, and work toward them. Listen to everyone who wants to give you advice, even if you'll ignore most of it. You will find your path and you can help others find theirs.
Employment is a very personal experience. There will continue to be the need for employees, managers, executives, entrepreneurs, innovators, and solo agents. Look decades out into the future. Think about what excites you, what the essence of the value is that you bring today and how you might want to bring even more value in the coming weeks, months, years, and decades to come. Continue to learn, engage, and guide people forward, and you will have a long and exciting career.
Let's turn our attention to the future and where opportunity lies for broad sections of the population and general workforce.
The future is long – there is much work to be done
Being deeply involved in the evolution of data, analytics, and learning systems for over 30 years, I never seriously believed, and do not believe, that AI systems will outright replace most humans in the realm of work.
The idea that people will be replaced by software across a wide range of industries in a short time period—causing despair, depression, and loss of motivation and engagement across societies—will not happen. There are people across the world who are prone to fearmongering and have an interest in furthering the dialog at the extreme end of the spectrum. Software that provides process automation and enables predictions will not replace entire industries in the short term.
In my first book, Analytics: How to Win with Intelligence, I wrote:
It has taken 50 years to completely automate transactional systems. It has also taken about 50 years to build out the first layer of information management systems, and we have not completed that ecosystem yet.
Looking 10 to 15 years in the future, we foresee having very sophisticated and automated modeling, data preparation, and model management systems. During that time, we will also continue evolving the math and analytical techniques used. For such systems, we have typically approached horizontal problems first – marketing effectiveness, customer loyalty, manufacturing quality, cyber security, and so on. It will take another 20 to 30 years to perfect these systems.
Somewhere in that time window, we will start to build vertical applications for specific industries, such as automotive, healthcare, pharmaceutical, energy, security, and telecommunications. This will be a long, complex process.
Thus, in total, we foresee between 90 to 120 years' worth of work before we complete our analytics journey. Obviously, we have much to do, but thankfully the work has been both interesting and engaging. 
Given the difficulty we have seen in the market for self-driving cars and related technologies, I stand by my prediction that we will not see widespread deployment of AI-based platforms that will cause significant job transformation and reformulation until 2150.
If you were counting on the creation, provision, and delivery of universal basic income due to losing your job to an AI system, you will be very disappointed. Best to keep your skills sharp and keep going to work each day. For nearly every advanced analytic system that my teams and I have built, we have worked with the subject matter experts and their teams after the implementation, and the employees are happier and are focused on higher-value work. Moreover, the humans often take credit for the better decisions being made that are either completely the work of the analytical model, or substantially supported by the output of the applications and/or models. No one ever complains about being on target more often through the augmented workflow of the human/AI collaboration. And the AI system never asks for credit, so it works out well for all involved.
In my personal experience, there is a significant amount of mundane and boring work to be automated away, and a corresponding amount of critical thinking and higher-level decisions to be made every day. These higher-level decisions remain in the remit of human cognition.
Learn and leap
"The lessons of technological innovation remind us that progress always entails thinking the unthinkable and then doing things that were previously impossible," Tim O'Reilly, the founder and CEO of O'Reilly Media, says in chapter 15 of his most recent book, WTF?: What's the Future and Why It's Up to Us. That's why he's optimistic that technology will augment, not replace, jobs. But, he says, "learning will be an essential next step with each leap forward in augmentation." 
In their new book, Augmented Intelligence: The Business Power of Human–Machine Collaboration, authors Judith Hurwitz, Henry Morris, Candy Sidner, and Dan Kirsch define augmented intelligence in this way:
…beyond artificial intelligence, there is augmented intelligence, which can significantly transform how we can leverage knowledge, artificial intelligence (especially machine learning), and various tools that support advanced analytics. So, what is augmented intelligence? Augmented intelligence is an approach that uses tools from artificial intelligence to perform well-defined tasks, such as those that are part of decision making. But for augmented intelligence, the human works in collaboration with the machines. Humans need to evaluate the results of automated tasks, make decisions in non-routine situations, and also assess if and when the data must be changed due to changing business needs and demands. 
We are on an evolutionary journey in developing and deploying artificial intelligence systems. A few facts will help set the stage as to the global effort in developing systems with advanced analytics and AI:
According to Evans Data Corporation, there were 23 million software developers in 2018, this number is expected to reach 26.4 million by the end of 2019 and 27.7 million by 2023. 29% of developers worldwide were using some form of AI or ML as of 2018 and an additional 5.8 million are expected to start using AI or ML within the next 6 months. 
The number of people who are developing these systems may seem like an overwhelming army of people toiling away, in every company imaginable, to automate away every job possible. This is not the case. While there are a significant number of developers working with AI and the number is growing, the majority of those developers are just beginning to experiment with AI-based technologies. And while AI and advanced analytics as a general topic garners an outsized amount of coverage from the press and pundits, we are in the early stages of a long journey.
Let's recap the main points discussed up to this point. AI will have an effect on jobs—it will actually, in all likelihood, create more jobs than it eliminates, but we need to be aware that these new jobs will be different and require higher-level skills than the jobs that are replaced. AI will augment and extend existing jobs. The idea that AI will eliminate jobs is overblown; in most cases, AI will remove the mundane aspects of work and allow people to focus on elements of work that call on more creative and subjective skills. Jobs have been evolving since the creation of work. AI is just another factor in that continuing evolution. AI is different and it has the potential to drive wide-ranging changes, but it is just another factor in the evolution of work.
To bring this process of job evolution into sharper focus, let's outline a recent example. One of my teams built a forecasting application, which we will discuss in greater depth in Chapter 8, Operationalizing Analytics – How to Move from Projects to Production. In relation to the operational staff who were employed to update and run the previous version of the forecast in a spreadsheet-based system, all of them were rendered unnecessary by the new forecasting application. Did all those 30+ people lose their jobs? No, they did not. Rather than having the staff manually obtain data, clean data, and load the data into a spreadsheet, the employees were retrained to be business analysts.
They went from being spreadsheet managers to analysts. Their job composition changed from 80% data management to 75% business analytics. Their new jobs are harder to learn, but pay more, are more secure, and provide for more job advancement and mobility. Now, the challenge is for the employer to retain them, given their newly acquired analytics skills. This is a much better place for the employees to be in from a career perspective, and the employees are more valuable to the employers. Everyone wins.
Now that we have discussed the potential impacts of AI upon jobs, and how those jobs can evolve into more secure and fulfilling jobs, let's consider another area where AI could bring major changes: the global education system.
AI in the education system
Over a decade ago, I was asked by the university where I did my undergraduate degree to review their new curriculum for teaching students the skills needed to be a business analyst. My input centered on the fact that there were no courses to impart communications skills and emphasize teamwork in the program. Reviewing the current course catalog, it appears that the suggestions I offered were added and remain in the program, and now, Ferris State University offers a program referred to as Data Analytics. 
I have spoken with administrators and staff members at the University of Illinois, Oklahoma State University, and the University of Michigan about data science, teaching data science, and preparing students for the changing world of data, data science, advanced analytics, and artificial intelligence. Clearly, my sample is small and limited to the university staff that I have been able to meet and talk with personally. What I have seen and observed is that teaching data science is being done most effectively and creatively outside the colleges of engineering. Engineering, unfortunately, has, in general, been slow to respond to the opportunities provided by AI.
The engineering curriculums are time tested and proven to produce graduates who excel in the disciplines required to be a successful engineer in the chosen field of study; fields like chemical engineering, civil engineering, mechanical engineering, electrical engineering, and many more. However, to be successful at the global, societal, national, and company levels, we need more qualified professionals than all the engineering schools in the world could produce each year.
Michael Webb of Stanford University, when talking about the need for universities to broaden their ability to produce well prepared graduates who can work in the fields of advanced analytics and data science, remarked:
New technologies create winners and losers in the labor market. They change relative demands for occupations, even as they improve productivity and standards of living. Understanding these distributional consequences is important for many purposes. For example, it allows policymakers to design appropriate education and skills policies, and helps individuals make good choices about what careers to pursue. 
The US and global educational system changes slowly, but it does shift according to the market. I have been working with New Trier High School for over a decade. Working as part of the advisory board, in collaboration with Jason Boumstein, we have reviewed and brought in courses related to engineering and artificial intelligence.
Tom Finholt, Dean and Professor of Information, School of Information, at the University of Michigan, has taken what was a program focused on library science and transformed the offering into a curriculum that focuses on teaching, training, and preparing students to be leaders in the fields of data science, UX design, and more.
Tom has moved the program away from the previous paradigm, that is, of forcing left-brained students to memorize and execute technical strictures and structures in a rigid and driven manner.
I experienced the old style of teaching in my undergraduate and graduate programs. It does not work for large segments of the student population. I remember sitting in my Introduction to Assembly Programming class as a freshman. My professor said something similar to, "This class will be hard; 50% of you will not be here after the midterm. If you are a computer science major, you will need to take this class as many times as needed to pass in order to remain in the computer science program." No comforting sentiments in that speech.
…and the new
Dean Finholt and his team have created a program and approach that is more inclusive and diverse in how it attracts, evaluates, accepts, and educates tomorrow's leaders in data science. I am very excited to see how this new approach increases the number and quality of people entering the fields of data engineering, data science, advanced analytics, and related fields of work.
One of the more accessible methods for people of all levels of interest is online learning. The School of Information at the University of Michigan is using Coursera to broaden their reach. The course—Programming for Everybody (Getting Started with Python)—has enrolled nearly 950,000 students and has garnered over 73,000 reviews with an aggregate score of 4.8 out of 5. The instructor, Charles Severance (also known as Dr. Chuck), was tasked with creating an introductory course for people who were creative, open, and possess the ability to learn in a non-traditional manner. 
I do not see much interest from the colleges of engineering in augmenting or changing how they are teaching students to be data scientists. As I said, my sample is very small and quite limited, but I do see a significant amount of innovation and change in business schools, newly created colleges like the School of Information at the University of Michigan, and other schools and universities to attract new types of students in new and innovative ways.
If we continue to believe that the only way to provide society with the talent and skills needed in the future is to push people through rigid, and in some cases, outdated curriculums while sitting in lecture halls listening to graduate students, we will not deliver for the students, our communities, and the world in general.
The main points that I would like you to take away from this section is that our education system needs to evolve, and whether you are an adult returning to school, a high school student preparing to attend university, a parent of a student about to attend, or in the midst of attending university, you need to look closely at the educational offerings and curriculum of the school or program you or your child will be attending.
The old ways of teaching what are considered traditionally technical skills and imparting a body of knowledge, in some cases, are outmoded. This means they will not work for the broad audience that society needs to attend and graduate from university, with the skills needed to be successful in data science and advanced analytics.
The education system is changing, but as you or your child are entering that system, you need to be aware of what the current offerings are and how they can prepare you or your child for the realities of today and the future.
We have discussed how the educational system of today needs to change to serve a broader audience in the future. Let's now turn our attention to how people who are drawn to analytics and analytical thinking are unique, and how we can best support our colleagues, coworkers, and employees in our journey to deliver value to our companies and societies through data and analytics.
We are different
Over the past 37 years, as an observer of human behavior, I have learned a great deal about the people who are drawn to, and are good at, developing advanced analytics systems and environments.
Different in a good way
In the past five years, I have had firsthand experience of the failures I have discussed up close in companies as diverse as biopharmaceuticals, computer hardware/software, and research and consulting services. This is a widespread and ubiquitous problem.
I refer to myself, and the teams that I build and manage, as being "special snowflakes." I say this because it is evocative of the truth. In addition to being truthful, I say it because the differences we exhibit and embody are a very positive aspect of our personalities and the value that we deliver. The rest of the organization needs to know that these teams and individuals are different, and that difference can be, and is a source of power, change, and competitive advantage. These differences are not to be managed out or reduced; they are to be understood, nurtured, and employed for the greater good.
Each individual that I have hired over the years who has turned out to be a brilliant developer, programmer, data scientist, business analyst, system engineer, data engineer, or data architect has been an unusual or unique person.
The ugly duckling
I first encountered this dynamic over a decade ago. I was running the operations of a UK-based company. We had a staff member that did not get along well with the team, and not just one or two other people, but the entire staff in the location that he worked in. He was multiple levels down in the organization from my position, but it had been raised to my attention that the organization as a whole would prefer to, in British parlance, make this position and person redundant. To put it in the American vernacular, the team wanted him fired.
I met with him. We talked a few times and I realized that we had an emerging need in an area where this person had been taking night classes. He possessed early stage skills, a strong desire, and it appeared that he had an aptitude for the work that we needed to have done.
Working from home was not very common at the time, but his commute was inordinately long, and it was not productive to have him in the office. I proposed that he start to work from home, and that he begin, as a side project, to build what the company would need in the coming weeks and months, but not to tell anyone.
By this time, he had been transferred to me as a direct report. He and I agreed that he would continue with his existing duties, to see if he could and would be productive working from home, and we would collaborate on the side project.
He did not need to collaborate closely with anyone on his regular duties. He had to receive work, execute his work, and return the completed work to the team that delivered the input materials to him. He executed this work efficiently, effectively, and flawlessly. Also, he worked in the same mode with me on the side project.
When removed from the office and having his communication reduced to email and a few phone calls, his productivity soared. The idiosyncrasies that were the cause of the interpersonal difficulties were not an issue any longer.
One of the problems was that he would fall asleep at his desk in the office. People were put off by this and attributed it to him being lazy. As it turns out, he did his best work at night. I told him that I did not care if he worked all night and slept some of the day; as long as his work throughput, quality, and responsiveness were good, I had no issues with his schedule.
When the side project was complete, I presented it to the management team and included the staff members who were the most vocal proponents of this person being dismissed. The reviews were uniformly positive and unanimous that the work was good and that the group would be very happy to have this application, and platform, represent the brand and company in the market. The belief in the room was that this work was accomplished by an outsourced third party. When I announced that it was built by the staff member from his new side project while he worked at home, the reaction was a muted agreement that keeping him on staff and engaged was the right course of action.
After a couple years of this arrangement, he resigned. He had developed a small, one-man consultancy, offering his services to companies in the US, UK, and Europe. He worked from his house and was making significantly more money and seemed the happiest I had ever seen him. He went from being despised and derided to being a successful employee and eventually an entrepreneur.
He sent me an email a few years ago thanking me for seeing what others had not: his passion, skill, dedication, and drive to be a contributing member of the team. He was an outcast. He did not know how to ask for an environment that would allow him to flourish. And rather than looking for a way for him to be successful, the rest of the organization went about shunning him and making him feel incapable and unprofessional for how he looked and behaved.
This wasn't the only brilliant individual I've encountered during my time working in the world of business and analytics.
A diamond in the rough
I inherited a brilliant analyst when I took over a business unit of a large technology company. This person came to me after my first all-company meeting. In that meeting, I told the entire global operation that anyone could come to my office at any time to discuss any topic. I also explained that I came into the office early and usually caught up on email, communications, and projects before other team members arrived, but that I was open to having impromptu conversations at that time.
This person came in and explained that he had started out as a social worker, helping people through challenging situations in their lives. A few years before this conversation, he decided that he wanted to pursue a master's degree in social work. As part of his studies, he was required to take a class in introductory statistics. It turned out that he was a natural at math and statistics. He was unaware of this gift and talent. He explained that he had lived in the town where the company maintained its headquarters and that he had never travelled very far, but that he wanted to travel as much as possible.
I smiled and said, "Be careful what you ask for, we may able to give it to you."
Over the next year, this young man proved to be an invaluable staff member, due to the fact that he listened carefully to clients in all industries and of all levels of seniority and expertise, from senior executives to data scientists and business analysts who were both prospects and clients. He could translate what they described and needed into not only system specifications, but working prototypes quickly, easily, and accurately. He was amazing at his job. We had him on a plane each week, and he loved it. He travelled to Japan, Western and Eastern Europe, Canada, Mexico, and more.
His unique combination of empathy, math acumen, verbal, written, and presentation communication skills, comfort with uncertainty, lack of ego, and pure joy in helping people made him exceptionally successful. Of course, all of this comes with a unique set of personal needs and personality traits, but those idiosyncrasies made him who he is. He continues to be a valued employee, a reliable person, and a contributing member of the team and his community.
In both the preceding vignettes, the majority of managers would not have seen the synthesis of skills and traits as being diamonds in the rough. When they both came to me, they had been sidelined and pigeonholed as a "type" of person. That was wrong, and it was a disservice to both and the value that they could and did bring to the respective companies and clients.
- Optimistic yet skeptical
- Intensely curious
- Mostly introverted
- A combination of left and right brain orientation at the same time
- Prone to perfection
- Social, but reserved
- In some cases, they appear to exhibit a lack of focus or possibly too much focus
Management by a non-practitioner or novice typically ends in failure. Managing a team that is building advanced analytics environments is not the same as managing a typical information technology project. Many firms fail in the organizational design phase of building an analytics team. Organizations and managers often fail to realize that managing a group of analytics professionals is more similar to managing a group of creative professionals than it is to managing a group of programmers. We will talk more about this in Chapter 2, Building an Analytics Team.
A common mistake
The great majority of non-technical people, and I include corporate executives in this category, think and believe that a technical resource is the same regardless of whether they are developing transactional systems (for example, Enterprise Resource Planning (ERP), Customer Relationship Management (CRM), and so on), business intelligence, dashboards, or artificial intelligence applications.
Executives, across a wide range of industries, tend to think and act as if any technical resource is interchangeable with any other. And part of this problem stems from the fact that they see technical people as resources. They do not understand, and for the most part, do not try to understand, the nuances of the various differences between the skillsets, motivations, and intrinsic interests of individuals who possess technical talent. This mindset and view is patently false, and it is one that causes companies around the world to waste millions, if not billions, of dollars each year.
A partial cause of this situation is that the technology function is a relatively new addition to the corporate structure. Think of accounting, manufacturing, sales, and distribution.
These functions have been in organizational structures for millennia. No competent executive would say that we can take a top performing sales professional and put them in accounting. It sounds ludicrous to even say it.
But, given that the technology function has been part of the corporate environment for less than 80 years, we hear nonsensical statements like, "We can take the developers who created the CRM system and have them build the artificial intelligence system for predictive maintenance." This is a ridiculous idea that is bound to fail, but I have had senior executives, across several industries, say this exact thing to me and believe this is a reasonable statement or question. They truly believe that they know what they are talking about.
Part of this lack of understanding comes from the fact that senior executives know that the technology function is relatively new, and it is different, but they do not want to know how or why it is different. They are willing to delve into the intricacies of international tax policy or transfer pricing, but ask them to dive into the details of how an artificial intelligence system will revolutionize manufacturing or supply chain operations, and that topic is not worth the intellectual energy to understand at a deep and functional level. That is for the "technical people."
This problem is not limited to a lack of understanding of the composition of project teams, but also a lack of the skills, experience, and expertise needed to manage teams that are tasked with building advanced analytics systems. Again, senior management sees a person who has managed the implementation of an enterprise system like a business intelligence environment and/or application, and they assign them to build a self-healing supply chain management platform. A grave error.
These mistaken beliefs and the decisions made about hiring, organizing, and undertaking advanced analytics projects dooms a substantial percentage of efforts to fail before those projects have even started. And the disappointing fact is that the executives who have made these mistakes have no idea that they have set the project on the road to failure, even before any funds have been spent or a single person hired. And at the end of the failed projects, those executives will lay the blame at the feet of the technology, or the technical team or the outsourcing partner or all of them, but little to no cause for failure will be attributed to their lack of leadership or understanding of what was to be undertaken or accomplished.
I see this today in companies that has decided to place advanced analytics teams in the individual functional and operational areas around the world. The idea sounds rational, and the approach appears to be justified, and it can be, but it takes a deep understanding of the business and the analytical applications and technologies that will be applied. There are precious few people in this organization that understand the triumvirate of business requirements, technical skills, and solution development/application. What has happened in this specific instance is that managers have hired people who professed to have, but in reality, do not possess the technical and business skills to competently approach the solution development process. The disconnect between the business operators, the newly hired data scientists and analysts, and the hiring managers have delivered almost no results in the set timeframe.
When leaders and managers allow this kind of rudderless environment to be built or to develop, data scientists resign and move on.
The hiring managers are confused as to why they couldn't manage these projects, just like they did other, seemingly similar, projects, and the executives grumble that the teams did not deliver the promised return. A powerful mix of failure, frustration, and resentment ensues.
It is difficult to see the problems clearly from the individual perspectives of the players involved, but when you take a view with a bit of distance, it is easy to see that the projects undertaken in this approach had a very small chance of succeeding.
Given that we have outlined where we can see pitfalls in the strategic direction set for analytics teams and groups, let's examine the optimal organizational structure and where the analytics function and team should fit into the overall company.
The original sin
Senior management decides that they need to enter the race to build applications that contain artificial intelligence (whatever that means to them).
They decide that since most of the discussion that they observe in the management journals, technology press, and the media in general has a flavor of technology or comes from a geographic region known for technology, they will put the advanced analytics team under the technology function managed by the Chief Information Officer (CIO), Chief Technology Officer (CTO), or worse, under the Chief Financial Officer (CFO). Senior management has taken the first step to failing.
Let's take the case of placing the advanced analytics team in the technology department, or under the CIO, first. It may take a year or 18 months, but the advanced analytics and artificial intelligence team will fall prey to the information technology mindset.
Advanced analytics projects are not, for the most part, linear. They are iterative, marked by exhilarating successes, and punctuated by dead ends, missteps, and disproved theories. Most information technology professionals, while intelligent and mildly curious, do not have the interest or fortitude for the iterative or recursive nature of advanced analytics projects.
Also, information technology professionals typically do not have the skillset to develop artificial intelligence applications. Staff members in information technology functions have evolved over the past 30 to 40 years, from developing bespoke systems to installing systems, managing customizations, and configurations. Information technology professionals are more oriented to vendor and project management than they are to core development. There is no judgment in that statement; it is a proven fact.
The information technology function is a support function—a cost center, not a creative function—that lives by project plans and has a reticence or visceral fear of failure or being late or over budget. Information technology teams have been treated like a support function for the last 30 to 40 years, and the teams that work in the information technology department act accordingly.
Advanced analytics teams, at least high-performing analytics teams, are seeking to solve one or a series of challenging problems. The pursuit of the solution is the goal, not adherence to a budget number or delivery according to a preset date. The optimal solution, the source of competitive advantage, is the objective.
Placing the advanced analytics team under the accountability of the CTO is better than under the CIO if the CTO has an innovation charter within their remit. As an example, the CTO's remit may include finding ways to productize unique data generated by operations.
This would likely be an area that the CTO would have an interest in driving, while also having the connections and political capital to enact it successfully. One of the downsides would be getting the CTO team to engage with the functional areas that the CTO is unfamiliar with, like IT operations, sales, marketing, HR, finance, facilities, and so on. The CTO will typically be tasked with creating new approaches to existing challenges. If this is the case, then it is possible that the AI team will have a fruitful run as part of this functional organization, but success is not guaranteed.
Drawing the reporting lines of the advanced analytics team under the organization of the CFO is worse than putting the team under the CIO. At least under the CIO there is a distant history of technologists building and deploying solutions. Under the CFO, the entire mindset is process orientation and cost containment. Not exactly the wide-ranging, creative mindset needed to develop novel solutions to drive competitive step change in an organization and industry.
Again, it may take a year or 18 months, but the advanced analytics and artificial intelligence team will fall prey to the finance organization mindset. The finance team is a very focused group, and they should be. They need to manage the financial flow, systems, and reporting that make the company operate efficiently and effectively. As the finance team gets squeezed to do more with the same staffing or less, they will look for relief. They will look to cut heads in the advanced analytics team to hire more people to run the day-to-day processes of finance and accounting.
The objective and goals of the advanced analytics team and group will rarely, if ever, align with the corporate goals and objectives of the finance team. And you know what happens when there is a misalignment of goals and objectives with the management above any team: that team gets their headcount and budget cut or held at a level that does not enable expansion.
We have discussed where not to put the analytics leader and the advanced analytics team, so now, let's outline the optimal place in the organization for the analytics function.
The right home
As I have said, the most successful advanced analytics teams are creative groups staffed with talented, motivated, curious people who can convert business discussions with subject matter experts into analytical applications and solutions that can drive operational change on a daily basis. The analytical teams that realize the most success have wide-ranging mandates to drive practical and pragmatic change resulting in competitive advantage.
Where in the organization are the senior executives whose mandates encompass this arena?
The best organizational home for the advanced analytics team is reporting directly to either the Chief Operating Officer (COO) or the Chief Executive Officer (CEO).
If a COO exists, it would be rare to see this group under the CEO, unless the CEO is younger, ambitious, and engaged. Recently, there have been a handful of Chief Analytics Officer (CAO) appointments reporting to the CEO. This is typically a move to illustrate where the CEO wants to place emphasis. It will be interesting to see how long this type of reporting relationship remains in place, before evolving into a new form and structure.
Working directly for the CEO can work, and it is one of the optimal reporting structures for the CAO and the advanced analytics teams, but it is often difficult to gain time with the CEO to ensure alignment and focus. To be clear, reporting to the CEO is the best direct reporting relationship the CAO can, and would, want to have, and if the CEO prioritizes the relationship with the CAO and publicly funds and supports the mission of the advanced analytics team, then this is the best possible organizational structure.
Having the CAO report to the COO is the next best reporting structure to have in place. The COO has the corporate functions under their control, and can direct the functions to collaborate with the CAO and the advanced analytics team to examine processes, data, and more to drive innovation and change.
Continuing down the senior management structure, the third best place for the CAO to report is to the Executive Vice President of Business Development and Strategy. This role typically owns mergers and acquisitions, strategy, and corporate development. Therefore, the mission to drive innovation typically resides in this group. Given the amount of change that the CAO and the advanced analytics team will drive, they need to be reporting to a corporate change agent with the organizational power to direct the functional groups to engage and collaborate.
We've discussed the implications of having a CAO and an advanced analytics team, and we have talked about appropriate environments in which such a team and leadership might thrive. This brings us to the final section of this chapter. As we conclude our exploration of the operating context for, and overview of, analytical teams, we need to discuss the topic of ethics.
I considered writing a book on the implications, needs, and requirements of ethics in analytics, but I decided against completing that project. Why? First, there are several very good books that have been published on the topic. Second, ethics should be part of our mindset in everything that we do. Given that belief, I felt that it was better to imbue this book, and every book that I will write in the future, with a subtext of ethics. All through the content of this book, I will discuss and call out how ethics impact and inform choices, and how to consider the ethics of the decisions being made.
Society, as a whole, is concerned about the implications of advanced analytics and artificial intelligence. Given what has happened with social media and the bad actors involved in those platforms, there is replace with: grounds for concern which drives the need for consideration that all innovations have the potential to be deployed and employed for unethical purposes.
Researchers, experts, pundits, company executives, government ministers, legislators, and others have written and spoken at length about the potential value and operational downsides of governmental interventions like the General Data Protection Regulation (GDPR) enacted by the European Union (EU) and the California Consumer Privacy Act (CCPA). One fact is for certain: we have passed the point where self-regulation by companies is an acceptable approach to constraining corporate policy and behavior in the use of data, advanced analytics, and artificial intelligence.
Governments are stepping in and will continue to step in. If building a business on a fair business proposition and providing value to customers, consumers, partners, and society is not enough of a guideline for corporate executive and managers, there will be more guidelines and regulations forthcoming from multiple jurisdictions.
Let's proceed under the premise that everything we do and all the choices we make in relation to data and analytics have the best interest of the consumer, customer, patient, and all involved in mind. Let's also keep in mind that the further we stray from this core premise, the more likely it is that we will cause harm to our business, reputation, and the wellbeing and peace of mind of our customers, patients, and broader society.
The further afield we take our efforts away from being ethically aligned, the harder the fall will be when the correction comes from either internal or external forces. Be assured that we will be forced to explain our actions and efforts at some point in the very near future.
Transparency, ethical actions, and honesty will be our guiding principles.
In this chapter we have started our journey of understanding the market and organizational context in which high-performing analytics teams live and operate.
We have touched on the perception, reality, value, and concerns surrounding AI and its implications for jobs and careers, which are mainly positive in the long term and disruptive for some in the short term.
We have discussed how analytics professionals are not your typical employees and should be evaluated from a results perspective, and not on the basis of their supposed interpersonal failings or challenges. This aspect of hiring and managing high-performance analytics professionals will be replace with: thwarted in the current work environment of hyper political correctness, but should also be helped by the move to incorporate diversity and inclusion in our organizations.
We touched on the optimal and less than optimal organizational structures that should and shouldn't be used to house and grow an advanced analytics team and its functionality, and we wrapped up by touching on the all-important topic of ethics.
We have set the stage for our wide-ranging discussion of how to be successful as an analytics leader. Thank you for coming on this journey with me.
Now, let's move on to discussing the topic of how to hire and build a high-performing analytics team.
Chapter 1 footnotes
- Will robots steal our jobs?, August 20, 2019, Mike Colagrossi, https://www.weforum.org/agenda/2019/08/the-robots-are-coming-but-take-a-breath
- Polyani's Paradox, https://en.wikipedia.org/wiki/Polanyi%E2%80%99s_paradox
- Rob May's thoughts on startups, angel investing, and becoming a venture capitalist, http://coconutheadsets.com/
- Gartner Says By 2020, Artificial Intelligence Will Create More Jobs Than It Eliminates, December 13, 2017, Rob van der Meulen & Christy Pettey, https://www.gartner.com/en/newsroom/press-releases/2017-12-13-gartner-says-by-2020-artificial-intelligence-will-create-more-jobs-than-it-eliminates
- Robots aren't stealing all our jobs, says the World Bank's chief economist, January 11, 2019, Sean Fleming, https://www.weforum.org/agenda/2019/01/robots-aren-t-wiping-out-jobs-yet-according-to-the-world-bank-s-chief-economist/
- You Probably Need a College Degree to Get a Factory Job Now, December 10, 2019, Kaitlin Mulhere, https://money.com/manufacturing-jobs-college-degree/
- Analytics: How to Win With Intelligence, January 7, 2017, John K Thompson & Shawn Rogers, https://www.packtpub.com/data/analytics-how-to-win-with-intelligence
- WTF?: What's the Future and Why It's Up to Us, October 10, 2017, Tim O'Reilly, https://www.amazon.com/WTF-Whats-Future-Why-Its/dp/0062565710
- Augmented Intelligence: The Business Power of Human–Machine Collaboration 1st Edition, November 1, 2019, Judith Hurwitz, Henry Morris, Candace Sidner, Daniel Kirsch, https://www.amazon.com/Augmented-Intelligence-Business-Human-Machine-Collaboration-dp-0367184893/dp/0367184893/ref=mt_hardcover?_encoding=UTF8&me=&qid=
- How many software developers are in the US and in the world in 2019?, December 27, 2019, https://www.daxx.com/blog/development-trends/number-software-developers-world
- Ferris State University, Course Catalog, http://catalog.ferris.edu/programs
- The Impact of Artificial Intelligence on the Labor Market, Michael Webb, Stanford University, November 2019, https://web.stanford.edu/~mww/webb_jmp.pdf
- Programming for Everybody (Getting Started with Python), University of Michigan, Coursera, https://www.coursera.org/learn/python, Dr. Charles Russell Severance, https//www.coursera.org/instructor/drchuck and https://www.dr-chuck.com/