Introduction

This book is intended for senior managers and executives who are contemplating or have made the commitment to hire and manage a team of inpiduals with the stated purpose of designing, building, and implementing applications and systems based on advanced analytics and artificial intelligence. If you believe that this objective can be accomplished in a year or less and can be achieved cheaply, do not buy this book. You should stop reading, cancel your current plans, and save your money.

Making a commitment to drive your organization to a higher level of effectiveness and efficiency is what you are doing. Just as if you were deciding to build a state-of-the-art factory or if you and your team were entering into a completely new market or geographic region of the world – if you do not have a multiyear view, then you should seriously reconsider undertaking this journey.

In this introductory section, we will walk through the process of succeeding in the analytics journey from beginning to end. We will assume that you are just starting to consider building a team, selecting projects, succeeding and failing at those projects, learning from those failures, moving from a development and test mode environment to a production mode environment, evolving from manual or traditional digital processes to data-driven, analytically enriched processes that improve with time, and undertaking the related and resulting organizational change management initiatives that are required to actualize the value realization that you planned for when you started this journey; and, of course, managing the executive expectations of the scale of investment, the scope of change, the speed at which the return on investment will be realized, and the realities associated with becoming an analytically driven organization. This set of steps comprises the "macro" process of an organization becoming analytically driven.

Becoming data and analytically driven

One of the mindset changes as well as the organizational process changes that is required to be successful in this journey is that by becoming a data and analytically driven organization, you at some point realize that the organizational change you seek is never "done." This process is evergreen and ever changing.

Along with the "macro" process of organizational and mindset change, there is a "micro" process of evolving and changing in response to the needs, wants, and desires of customers, patients, the market, the environment, suppliers, investors, stakeholders, and competitors. From the perspective of the middle of the processes as described, these processes at the execution level are usually described, organized, and discussed as projects. The larger overall process is typically made up of projects that focus on a specific objective or goals, but the overall process is dynamic and ever changing. If you, your leadership, and your organization want to be part of this continual evolution, then your organization, data operations, and analytical models and processes need to be set up and organized in a way that accounts for and reacts to the constant collection of relevant data inputs, updates and tests the models being developed and deployed, monitors execution and performance, and refreshes the analytical contents and models at the appropriate time and cycle.

One of the pitfalls that most organizations fall prey to is to think that the projects are the process – they are not. The projects are the execution mechanism. The projects are important for organization, management, funding, tracking, and reporting, but the projects are the trees. The forest is the organizational commitment to being analytically driven, the desire to continually improve, the interest in being open to new data and the evolution of the market, services, competition, consumers, patients, and stakeholders. The forest is the strategy to be a market leader. Do not lose sight of the forest for the trees.

An analytical mindset

For those readers who are analytics professionals, let's draw an analogy. Advanced analytics models are trained on data. The data represents the world or the subject area at the time that the data was collected. Once the model is trained and that model is accurately "predicting" the characteristics of the subject area as represented by the training data, the model is "locked." By locking the model, we end the training phase and we move the model into production.

The model ingests and examines data in the operational world and predicts the information that we are interested in. But we all know that the world changes and so does the data that is the byproduct of those activities. The models must be updated or retrained using current data to ensure that the models are generating predictions based on data that is as close to the current state of the world as possible. We "unlock" the model and train it again using new data. The model now predicts based on the new frame of information. The model continues to track the evolution of changes in the operational world through these cycles of training and production through its operational life cycle.

People are like analytical models, but sadly, the great majority of people lock their mental models and rarely, selectively, or not at all do they open those mental models to update their views to align with the new reality of the world. This is one of the primary reasons why people are called "out of touch" or clueless.

Let's, as a group, keep in mind that the world and all the phenomena in it evolve and change. Just because we were exposed to a set of norms, technologies, cultural constructs, and other conditioning when we were growing up, that does not mean that we cannot retrain our models to include new norms and activities. We do not need to throw away the old, but we can include the new and have a richer view, a more inclusive view of the world. Let's unlock and retrain our mental models frequently.

Thank you for indulging that brief detour. Let's now return to discussing and describing the "macro" process, or having an analytics strategy.

The essence of being analytically and data-driven is to be dynamic and guided by empirical evidence and measurable factors. Some of the most successful companies are organized and operate in this manner and have done so for years, but now the speed and pace of change have made this way of operating an imperative rather than a nice-to-have for firms that want to be at the top of their league tables.

Building an analytics team and an environment for collaboration

You will need to hire a leader. You will need to empower and fund that leader to hire and lead a team. I refer to this team as the Advanced Analytics and Artificial Intelligence Center of Excellence (AA&AI COE). The COE leader and team will need to learn the broader organization and to build a network of collaborators, sponsors, and allies. This new network, which we will refer to as the Community of Practice, or COP, will span the entire organization and global operations. If your organization numbers in the thousands, the AA&AI COE will be less than 50 staff members within the first two years and the COP will be a few hundred in number.

The most recent COP that I built took approximately a year to reach critical mass and was made up of between 400 and 500 staff members spanning the globe, including every operational department in the company. I stopped counting after the first 14 months, but, in the first year, I held over 600 introductory meetings. To be clear, I did not meet with 600 hundred people, no, I met with well over 1,500 people – I had 600 meetings and some of the meetings were with 20 to 50 people.

I traveled to every continent and spent much of my time on the road and in discussions with executives, managers, and people who should be involved.

The primary objectives of those meetings were to:

• Communicate and convince executives, senior managers, managers, and inpidual contributors that they should collaborate with the AA&AI COE team and the other staff members in the COP
• Communicate that the AA&AI COE team was established to help them understand and develop analytics to drive their operations forward in the manner that they dictate
• Empower their staff members to join the COP and to join projects in conjunction with the AA&AI COE staff to develop analytical applications, models, and new ways of reaching higher levels of effectiveness and efficiency
• Let them know that we were not there to judge their ideas and current state of operating, but to help them see how data and analytics will help them reach and exceed their operating goals
• Improve employee engagement and remove the tedious parts of staff members' duties to enable those staff members to focus on the more creative aspects of their work that leverages their experience and expertise

Collaborators in the analytics journey

Let's define the taxonomy and naming of the collaborators that we will discuss and describe in the book and work with on our analytics journey. I will be referring to executives as sponsors, because they typically set the direction and control funding and staffing for their organization.

Also, I will be referring to managers as stakeholders, as they typically own the headcount that is needed to collaborate with the AA&AI COE staff. And finally, I will be referring to the staff members of the operational departments as subject matter experts. The AA&AI COE leadership, data scientists, data engineers, data visualization experts, and others cannot be successful without the full-throated support of sponsors, stakeholders, and subject matter experts. A transparent and trust-based relationship between all parties is crucial to our joint success.

I have explained in keynote speeches, fireside chats, articles, white papers, meetings, books, and more that AA&AI will change jobs, and in the process eliminate certain jobs and job content, but the process is to take the robotic work out of a given job and enable people to focus on higher-value work and tasks. Our objective is not to take jobs away from people and give those jobs to robots. It is to make work more engaging and fulfilling for people.

The new AA&AI COE and the COP network will need resources and funding. The executive leadership of the organization will need to publicly support the related data and analytics initiatives.

Selecting successful projects

At this point in the process, the analytics leadership and the AA&AI COE team can begin to discuss and select an operational area or areas to partner with to begin to evaluate options for analysis and possible operational improvement. The AA&AI COE will discuss the potential areas for improvement with the operational or line-of-business executives and teams. The discussion and selection of areas for examination and improvement through data and analytics sounds like it should be straightforward and encounter little to no contention in the process, but even in the most apolitical organizations, this is not the case.

Organizational dynamics

Executives and senior managers will span the range of behaviors, from exuberant and public support to actively attempting to block projects and progress.

Building the AA&AI function and capability in an organization will be seen by some as an opportunity to grab more budget or funding and to create a larger organization or empire. You, as the executive sponsoring or building this new capability, need to be aware of this dynamic and ready to evaluate the motivations and abilities of these managers. It is entirely possible that these existing managers may understand data and analytics and have the experience, expertise, and knowledge of how to be successful in the macro and micro processes described earlier in this introduction. If that is the case and you believe in them, then the organization has a head start in the process, but if they do not have the requisite experience and knowledge, and they think that the success that they have had in other, possibly only tangentially related, areas or projects will propel them to success in this venture, then your efforts will have a tumultuous start.

The premise of this book is that undertaking the journey to build an organization or organizational function/capability to be analytically and data-driven is unique and different in many respects. In my personal experience, it has been proven to be the case numerous times. Handing this process to people who have little to no direct experience is a mistake and it will prove to be so nearly 100% of the time. One of my animating and driving factors for writing this book is to help you avoid this fate and failure the first time, not the second or third time.

If this is truly the first time that an organization is engaging in building systems and applications based on advanced analytics and artificial intelligence, then the number of potential areas for improvement will be numerous.

There will be no shortage of opportunities to improve the business and realize significant returns that exceed most of the existing investment returns in the corporate project portfolio.

In this environment, one of the challenges will be to convince the executives and managers that the initial returns being calculated and communicated are real and can be realized. In this case, the hype is true. Systems and applications based on AA&AI will change all aspects of work and operations, if companies have the fortitude to hire and engage AA&AI teams, to let the new AA&AI COE team members engage with sponsors, stakeholders, and subject matter experts in the existing operational areas and lines of business, and the organization has the breadth of vision to know that being analytically and data-driven means changing how the firm operates in all aspects of the organization. If these conditions are true and exist in earnest, you are truly lucky and should set to work immediately. The stars have aligned, and your journey will, mostly, be smooth as you set off on your path.

If you are fortunate, and I have been fortunate in most of the organizations I have engaged with, you will have one or two, or maybe a handful, of executives and senior managers that will step forward and jump into the analytics process with full commitment. They will tell their teams that the AA&AI COE is to be considered part of their team and that they want to attack vexing problems and not to change only at the margin of operations, but that they want to change operations at the core. They want to use data and analytics as a competitive tool to best the competition.

These are the people who will be promoted in 18 to 24 months based on the risk they took with you and your team. I am happy to say that over the past 3 decades, my teams have helped multiple people move up in their careers. Those sponsors have gone out on a limb for us and we have delivered for them, their teams, their organizations, and, in some cases, their higher aspirations.

You will be supported by the sponsors and pioneers that we just mentioned, while you will be opposed by the naysayers, Luddites, and those that want to abscond with your funding and team. After you, your AA&AI team, and the pioneers and their teams deliver groundbreaking functionality, operational improvements, and significant returns in the first year to 18 months, a portion of the opposition will convert to supporters. These are the people who jump on the bandwagon or get in front of the parade once success is guaranteed. Welcome them for the time being; if the tide turns in the future, they will turn against you in a heartbeat. They are not allies, but can be useful when the time is right. The naysayers and Luddites will always oppose you, your team, and your agenda. Be polite, be kind, and roll over them with success.

To close the discussion of organizational dynamics and politics, if the organization numbers in the thousands, spans the globe, and is committed to improving through the use of data and analytics, your team of a handful of people cannot deliver on every possible area of improvement. There are two areas of augmentation of your capabilities that you should support and actively promote.

Competitive advantage or simply staying competitive

First, you should consider outsourcing certain projects to competent, capable, and proven third parties. The projects to be considered are those that others in your industry have completed and are now considered tables stakes to be competitive at the new level of efficiency and effectiveness that the industry operates at or that the market, customers, suppliers, and patients demand. Projects like inventory management, supply chain efficiency, or designing servicing maps for optimal territory coverage by a sales team – these are projects that have been successfully executed across numerous industries with well-known and published success stories.

Find an experienced third party with a long track record of success and outsource the project. Of course, you need competent team members to manage the project, but you do not need analytical professionals to manage this relationship and process. To be clear, you cannot wash your hands of the project either. The project will need oversight and analytical validation, which you and your team will need to provide, but daily supervision is not required on your part.

Second, you should support the operational areas that want to invest in business intelligence, descriptive statistics, and small-scale predictive analytics. Your team cannot do it all, but your team can help these functional areas hire the right staff members, undertake initial projects, and consult on predictive and prescriptive applications in the future. Helping the organization build a broader and deeper capability is part of your responsibility and it will build a network of supportive team members who aspire to grow in their skills and abilities and may be good candidates to join the AA&AI COE in the future. It can only help your cause and ease your journey to build an ecosystem of talent that the organization pays for and nurtures that could become the future talent for your team.

The core collaboration/innovation cycle

Once sponsors have committed their teams to the process of improvement through data and analytics, and candidate areas for improvement have been selected, then the AA&AI COE team and the functional team (that is, the sponsors, stakeholders, and subject matter experts) can begin to analyze the operational area, the processes, the data, the organizational resistance to change, and the feasibility that the required data exists today and in a historical form. The AA&AI COE team can explain to the functional team the analyses that will be experimented with and undertaken and the analytical pipeline and models that will be developed.

The functional team will need to understand the process changes, the new mode of operating, and the resulting changes to their personnel needs and daily functions.

After a common understanding of the area to be analyzed and improved through data and analytics has been developed, the functional team and the AA&AI COE team can begin to gather data, build pilot environments, and discern whether the hypothesis developed to this point in the process is possible and probable. There is a non-trivial chance at this point in the process that the hypothesis will be proven to be incorrect and/or the data required will not be available in the quantities and historical depth needed to build reliable models. There are several reasons why the pilot could prove that this path will not drive measurable or significant business value or not be technically feasible.

Let's assume, for the sake of our discussion, that the pilot project delivers measurable results and it appears that a robust analytical model (or models) and a revised and improved operational process that includes the analytical pipeline and model(s) can be designed and implemented. In addition to the data being available, we have the legal and ethical right to use the data for the stated purposes, the functional team and leadership remain in their roles, and the senior and executive leadership remain committed to the effort. In short, all the organizational contextual factors remain constant or at least supportive of the effort we are undertaking.

With the pilot project complete, the probability of a prototype being technically successful is reasonably possible, the organization remains committed, and funding continues, now the AA&AI COE team can begin to collaborate with the functional team to build a full-scale prototype system or application to prove that the system will work as designed and deliver reliable analytical results to the operational team members who will use the production output to change decisions and reach a higher level of productivity, efficiency, and effectiveness.

Once the prototype has been built and tested, then it is time to begin to talk about how to implement the next iteration of the data flows, analytical models, governance systems, process changes, and end user interaction systems to realize the benefit of all the work that has been completed to date. It is very difficult to state this next fact, and even harder for most readers to believe it, but this is where most efforts fail. We will explore the numerous reasons why many analytic projects fail at this point later in our discussion in Chapter 8, Operationalizing Analytics – How to Move from Projects to Production.

Let's assume that the process does not fail, so now is the time to build data feeds for production systems, prepare the routine and regular processes to move from analytical modeling in test environments to inserting and refreshing analytical models in production systems, and prepare functional teams to understand the changes that will be required from them to support the newly modified production processes.

Focusing on self-renewing processes, not projects – an example

Perhaps a concrete example can ground our discussion to make it clearer and more understandable where we stand in the overall process and the pitfalls that face the combined analytical and functional teams.

Let's use the example of retail store site selection. Before the widespread use of data and analytics, organizations had site selection teams that performed research on the proposed region, markets, and neighborhoods; visited prospective locations; scouted sites; and spent time talking with local business people, local governing bodies, landlords, real estate agents, brokers, and others. They compiled briefing books on the various options and made presentations to management regarding their process and the locations that the site selection team felt were the best bets. This process could take weeks to months to possibly years.

Why is this traditional process not optimal and why does this type of process drive urgency in a dangerous way?

Most of the data collected in this type of process is static and refers to a particular point in time. Revising the supporting analyses can take weeks or months due to the need to go back to source systems and people. There is a sense of urgency in the process, which is a good thing, but in many of these cases, the sense of urgency is because the analysis, the possible target location, and the conditions of the analysis have an expiration date.

What do I mean by an expiration date? Much of the work executed was executed as a project. The datasets were gathered at a point in time; the conversations were all relevant to the day, and maybe the weeks, in which they took place; and the prospective location may be of interest to numerous buyers for multiple purposes. The relevancy and timeliness of the data and the analyses derived from the data begins to age and deteriorate in value the day it is complete. The main objective of the project is to select a site. The selection needs to be made relatively quickly or the work that was done will become out of date or stale. If too much time has elapsed, the data gathering, creation of the analyses, and the evaluation of the potential sites will need to be done again to ensure that the conditions as understood are still valid. Urgency for the sake of ensuring that the conditions remain as outlined in a report is not a good reason to move quickly.

What does the new process look like when augmented with data and analytics?

The overall process for the most part remains the same from a structural perspective. The site selection teams perform research on the proposed region, markets, and neighborhoods, visit locations and scout sites, and spend time talking with local businesspeople, local governing bodies, landlords, real estate agents, brokers, and others.

But the differences are as follows:

• The new store site selection application is built to automatically receive updated data from all relevant sources when those sources are updated. All data is always up to date and ready to use without additional effort from the AA&AI COE team or the functional team.
• The first activity that a site selection team undertakes is to run a model that analyzes all relevant data on current income, demographics, unemployment, similar store sales for the competition and for industry comparators, same-store sales in the organization's current operations, and any other relevant factors that the company has decided have an impact on the success of new locations. The site selection team will run this model for the entirety of the target country, not just the local area for the potential new store.
• The application provides the ability to simulate and optimize the entire network of stores and considers cannibalization of the company's existing stores by new stores, competitive market entry, and store closings. The effect of competitive and own network activity, such as pricing changes, marketing campaigns, changes in operating hours, and more can easily and quickly be modeled and examined.

The site selection team is now examining strategic store network considerations rather than trying to decide where to put a store. The cycle of iteration given the new application has a span of minutes rather than days or weeks. The site selection team can consider thousands of scenarios before they present a single plan.

Simulations and optimization work literally has no limits. The team can run millions or billions of scenarios if they choose to. This is how an artificial intelligence (AI) model or environment learns how to win at games like chess or Go, and in the new world of online multiplayer games, an AI application runs billions of games or simulations of games to determine how to win. This application for selecting a site is no different.

The site selection team, armed with the simulated optimal plan, will present to senior or executive management their recommendations. Now, if the team is truly confident, and they should be, and the senior or executive management is interested and engaged, and they should be, the meeting should not be a static presentation. The meeting should start with a presentation of the optimal plan but should evolve into an open discussion where varying objectives are proffered, discussed, and run through the site selection application in real time.

Senior management and executives will have contrasting and possibly conflicting views of what "optimal" means to them as inpiduals and to their functional areas of the company. The site selection team, supported by the AA&AI team, should be able to sit with the senior managers and executives and model the varying scenarios described by the attendees, and the application should run in real time so that results are presented in seconds. This iterative and interactive process enables the group to explore ideas and scenarios and optimize for the objectives that are most important to the operating results of the company in that moment.

What is different between the two processes? Previously, the site selection team was picking a site. After the collaboration with the AA&AI COE team, the site selection team models overall network changes at the store and program levels for the company.

The site selection team can now consider direct and indirect competition, including ancillary market factors such as shipping, unemployment rates, and income levels, along with other macro-economic factors. The site selection team can run billions of scenarios and execute real-time collaboration with senior and executive management to optimize the overall objectives of the company. Together, they can model the entire network to examine the impact of changing the operational plan to determine where to expand the store network before making any physical or operational changes. Only a small change…

That is the overall process you are considering, or possibly have undertaken, or maybe taken a swing at and are recalibrating your efforts. It doesn't matter which of those conditions are true – this book is a guide for you. A guide to help you understand best practices, to learn about pitfalls, and to shortcut your way to success.

This process has been executed exceptionally well by some and incredibly poorly by others. In some cases, the same person or team has done both! I have done both.

Summary

I hope that after reading this introduction, your interest has been sparked and that you are intrigued about how to execute and drive the analytics process forward in your profession and in your organization. The process will not be quick, it will not be easy, and in some parts of the process, you will not enjoy it, but it won't be dull, it won't be boring, and you will always be learning. If you are a lifelong learner, and the chances are good that if you have read this far you are, a career in analytics is one of the most fulfilling careers you can choose.

People continually ask me something along the lines of, "Analytics is such a hot and current field, how did you find yourself in it so many years ago?" My response is typically something like, "I am a lifelong learner. I live to learn new things. Not just things about data and analytics, but I do love learning about those topics too. I love to learn about everything – ethics, physics, math, credit risk evaluation, failure rates of PC hardware, human behavior while in distress, pricing dynamics, volunteering rates, retail merchandising mix modelling, and more." I go on to ask, "What other career can you be working on a credit risk evaluation application in the morning and discussing the optimal flow of oil through a plant that manufactures mayonnaise in the afternoon?" Most of the people I am interacting with say something along the lines of "None that I know of." And they are right; no other profession feeds lifelong learning and curiosity like analytics.

For those who are insatiably curious about our world, the universe, all aspects of human behavior, and more, analytics is the place to be. Welcome to the journey. I hope that you find this book useful, valuable, and fun to read.

I look forward to interacting with you soon at an event, online, or over the telephone.

All the best,

John