Building Highly Effective Analytics Teams

John K. Thompson, Author of Building Analytics Teams and Global Head of Advanced Analytics and Artificial Intelligence

Currently, I am the Global Head of Advanced Analytics & Artificial Intelligence at CSL Behring, a global biopharmaceutical company.  Before CSL, I was an Executive Partner at Gartner, acting as a management consultant to market leading companies in the areas of digital transformation, data monetization and advanced analytics.  Prior to Gartner, I lead the advanced analytics business unit of Dell. 

In my decades of experience, I have noticed, examined, and analyzed the fact that corporations, executives, senior managers and managers, in general, have significant misconceptions about how to be successful in using data and analytics.

My recent writings, articles, and book have been focused on the optimal way to build an advanced analytics and artificial intelligence function and provides guidance and instruction on how to:

• build, manage and grow a core analytics team
• acquire and grow the requisite funding
• manage executive expectations
• move from projects to production processes
• deliver real operational improvements
• navigate the pitfalls in the processes listed above
• deliver bottom line results & improve the competitive position of an organization

Corporations, governments and academic institutions and the leaders, executives, and senior managers employed by them, are making significant errors in in their attempts to build advanced analytics and artificial intelligence capabilities.  These missteps are costing each organization millions of dollars and years of lost productivity.

Many organizations move into working with data and analytics with little to no understanding of what they seek to achieve.  When organizations do attempt to hire an analytics team or set up analytics projects without a clear understanding of what they are trying to do or accomplish, these are the most commonly-asked questions after the initial effort have failed and the recently hired staff has resigned…

• Why did we spend millions of dollars and fail?
• Why did the data scientists we hired all leave so quickly?
• Why did the functional business users and the data scientist have such fundamental misunderstandings?
• Why could we not translate our initial findings into operational improvements?

Typically, when embarking on a new endeavor where you and your colleagues have little to no knowledge of the path to success, it is best to start by asking a few questions that will help in setting the direction that you and your team should undertake.  These are the questions that I see managers and executives asking before they begin their efforts to hire and build an analytics team…

• Where can we find someone to hire to lead our analytics team and initiatives?
• What are the qualifications and experiences best suited to start an analytics team and function?
• Are analytics and transformation related?
• What kind of education does the new leader need to be successful?
• We can’t find enough qualified people, where do we look for a broader group of people?

It is important for all organizations – enterprises, small businesses, academic, research organizations, and others to understand that an analytics team is not the same as a technology team. 

This is especially hard for organizations like pharmaceutical companies that are dominated by scientists and a majority of staff have science-based educations and backgrounds.  The view held in many of these science-based organization is that data and analytics is based on technology, so it must be just like a technology project or team.  This is a huge mistake and leads the majority of organizations down the wrong path.

Analytics teams are not technology teams.  You cannot take a technology team and convert them to an analytics team.  It will not work. 

Technology teams are staffed with people who think of linear productivity and are oriented primarily toward efficiency in executing a project plan. Of course, there are many technology professionals who are much more than the short description just given, but when technology teams are surveyed, examined and analyzed, one of their primary driving forces is effective execution.

Analytics teams, at least, highly successful and effective analytics teams, are not driven by effective execution.  Again, effective analytics teams do executive effectively, but that is not one of their core driving principles. 

Analytics teams are more like creative teams than technology teams.  Analytics teams are looking at processes and objectives with a view to create an analytically based solutions that bring together numerous systems, data sources, analytical approaches, algorithms and more, to develop innovative solutions that will deliver market changing insights and results.

Analytics teams need to be managed in a way that gives each data scientist the room to explore new and novel ideas and concepts.  The data scientists need to know that they can try new approaches and avenues and that they can fail, without rebuke or retribution.  They need to know that when they arrive at a dead end, that is ok.  Finding that something does not work is valuable in and of itself. 

Paradoxically, this is hard for pharmaceutical and science-based organizations.  It is puzzling that organizations such as pharmaceutical companies who are based on the premise that pure research and development will not succeed 100% of the time are some of the companies that find it hardest to understand that not all data science efforts will produce useful results and outcomes.  I hope to examine this dynamic more in my future work.

Knowing that not all data science projects, and sub projects will produce useful results, how does an analytics leader keep individual data scientists and the entire data science team delivering consistent and useful results to the organization? 

The answer is to assign each data scientist a personal project portfolio (PPP) that they own and manage.  Each team member on the data science team should have a PPP that they are responsible and accountable for delivering.  Each PPP consists of major projects, those projects that take between a few months and a year to complete, and minor projects, those projects that take between a few days or a few weeks to complete.  It is up to each data scientist to manage and deliver every aspect of each project on time and with the level of quality and utility that they outlined explained in the project charter for each project.

This concept of a PPP works well when the analytics team is structured as an Artisanal Data Science Team.  An Artisanal Data Science Team is one where each data scientist is responsible and accountable for all aspects of each project.  The data scientist writes the charter, collaborates with subject matter experts, presents to relevant managers and executives, and works with all parties to move the analytical project into production.

This approach is predicated upon the fact that an Artisanal Data Science team is staffed with highly qualified, high experienced team members who are motivated and driven to engage in the work of analytics and data science and want to make a difference in the organization.  This is how I have built and manage the advanced analytics and artificial intelligence team at CSL.

Of course, there is another approach to building an analytics team.  I refer to this approach as a Modular Data Science Team.  This approach is more akin to a production line where each small group of specialists work on one specific step in the analytics process.  One team or individual may be responsible for data acquisition, the next group works on data integration, the next on featuring engineering and the process continues to move through specialized groups or individuals until you have a completed process.

The Modular Data Science team works well if you want to have a larger team working on repetitive tasks where you can swap out staff in any role quickly and easily.

It is possible to have a hybrid approach as well.  I have taken a Modular approach to data acquisition and integration functions and then handed the results to a Artisanal team that works on the more intricate portions of developing predictive models.

This approach gives the organization and analytical leader great flexibility in how to hire, manage and develop the analytical team and function.

Analytics and the strategic use of data to examine and understand how to manage a process, function, department, division, and company is becoming the state of the art in leading companies.  Research has shown that market leaders are building and deploying advanced analytics and artificial intelligence based applications and models across their organizations and reaping outsized rewards from doing so.  This same research shows that those leading companies are creating a competitive advantage and distancing themselves from the early adopter companies.  There is concrete evidence that predictive analytics is a game changing technology and that it is being deployed and expanded today.

My view is that you and your teams should be talking with the executive leadership in your companies to determine the budget that will be set aside to establish an advanced analytics function in your company as soon as is practical.  While you are working out the funding process, you can begin to review the strategic challenges that you and your colleagues face today and begin to discuss how advanced analytics can help understand and overcome those challenges. Using data and advanced analytics will be part of every organization in the future.  The time to start your journey is today.