reread a chapter multiple times or look up individual terms in the index. The idea here is for you to come up with your own way to read this book effectively. And don’t forget to keep it nearby for quick-and-easy reference as needed while you work through your first few decision intelligence projects.
Part 1
Getting Started with Decision Intelligence
IN THIS PART …
Mining data verus minding the answer
Learn why math-only approaches are weak
Watching the details and missing the big picture
Discover the epiphany in the inverted V approach
Chapter 1
Short Takes on Decision Intelligence
IN THIS CHAPTER
Do you find yourself looking at a spreadsheet or viewing charts or gazing glassy-eyed at a fancy visualization that some bit of artificial intelligence magic has produced for you and wondering what you should do next? You’re not alone. Millions of other business and finance people are doing the same thing. So are legions of leaders and decision-makers in other industries.
While you’re trying to puzzle out which parts of those “actionable insights” being handed to you are in fact actionable and, if so, what action would apply, you’ve likely wished for something a bit more cut-and dried when it comes to determining what your organization would implement — and you certainly wouldn’t mind being considerably more certain about what's going to happen post-implementation.
Would your best bet in such a situation involve letting the miracle of artificial intelligence (AI, for short) make your decisions for you? Well, it turns out that AI isn’t so miraculous. In fact, an estimated 80 percent of all AI projects fail, where failure here is defined in terms of failing to deliver a measurable business value. That means most AI projects end up on the trash heap for leaning too heavily on the experimental side and being useless on the applied side.
It is painfully (and expensively) obvious that this strategy isn’t quite working out the way everyone hoped. An alternative approach is needed to make data more helpful and better aligned with consistently delivering business value. One such approach flips the model from data driven processes to decision driven processes. Known as decision intelligence, human and machine decision-making skills are combined with decision theory, decision sciences, and data sciences in a customizable mix that pins decisions to a precise and expected business value.
The concept isn’t entirely new — one of its oldest published mentions cropped up in 2002 in Uwe Hanning's scholarly paper “Knowledge Management + Business Intelligence = Decision Intelligence” — but it has evolved over time, incorporating long-accepted scientific formulas from several well-established sciences. This means its inner workings are well known and tested. Switching over to a decision intelligence approach is therefore no gamble — it's simply a supremely logical way for you to achieve the business outcomes you desire. Decision intelligence leaves little to chance, in either its own construct or the value it consistently delivers.
What differentiates one decision intelligence project from another is the talent and acumen of the decision makers at the helm. They make the recipe that cooks the business value into the process. And they decide when and whether to invite data and machines to the planning table.
Decision intelligence is highly agile and versatile. Decision makers can use it to make decisions either on the back of a napkin or with the help of the most sophisticated AI on the planet.
The Tale of Two Decision Trails
The business world has long been madly in love with the notion of being a data-driven enterprise, but it's also beginning to feel the pain of being in a bad relationship. Few actually want to break off their relationship with data entirely, mainly because most are loathe to ditch their significant investments in data, analytics, and related technologies. Add to this the fact that, for many, it would feel like a colossal failure and a huge embarrassment to fall short of becoming the data driven enterprise that all investors and stockholders expect these days.
Looking for a way forward, many started to ask themselves this question: “What can we do with all the data investments we've already made and already own in order to make better decisions?” In other words, folks realized that a rethink was needed rather than a redo. And many of those same folks decided that re-strategizing and restructuring how these same investments are used and aligning them with specific business impacts was the answer to the questions that had been plaguing them.
Pointing out the way
AI and data analytics no doubt deliver real business value in some use cases. They’re helpful when it comes to recognizing patterns in massive amounts of data and spitting out equations, scores, predictions, and estimates. The point is that such facts point to possible decisions but suggest none. (That's why I refer to the outputs from such tools as pointers in this section.)
These tools are also capable of automating certain decisions based on business rules that are determined and set by you or your organization. At its core, AI is automated decisions at scale. Traditional analytics must be integrated with automation software to cause an action to occur.
But before the various software — analytics, AI, and various forms of automation — begin their work in producing insights and automating your decisions, someone has to either program the analytics and automation software, and/ or train the AI. This group of data professionals often provide the interpretations of the outputs as well (usually as visualizations and/or automated AI-generated narratives).
In other words, people in specific job roles who do these tasks typically determine which insights — pointers, in other words — are accessible to other people in the organization who either use the software in a much more limited way or only view the results on dashboards to consume the information. Given the high degree of data illiteracy throughout organizations
