AI-Enabled Analytics for Business. Lawrence S. Maisel

are not predictive, and views of past data can lead to false negatives or positives of the future. Look at the image below [Figure A]…

The historical trend is essentially up. So, what's the next bar to follow? Up? Down? What decision would you make if you predicted up? What would happen if you guessed wrong? As seen on the chart below [Figure B], the next bar was substantially down.

       Figure A

       Figure B

      Visualization gives colors and images that intrigue the eye. But there is pretty and there is practical, and the two should not be confused—although they often are. Far too frequently, dashboards become an exercise in art vs. business. The rendering of a dashboard should be to make better decisions; so when viewing a dashboard, always ask, “Will what I'm seeing help inform me to make a better decision? What decision?” If the answer is not definitive, then the dashboard is art, not business.

      We like to say that AI and analytics can torture data until it confesses! The “confession” obtained from analytics, which applies mathematics on data, can better inform us about the future; and decisions are about the future! Consider, have you ever made a decision about the past? Well, no, other than to say that the decision you made when the past was the future turned out to be a good or bad decision. While this bit of time travel may be confusing, the point is that using tools that display data from the past is only part of the inputs needed to make decisions about the future.

      Therefore, it is important to distinguish that data visualization is largely a tool of reporting and displaying past data and information, whereas AI and analytics tools use past data to bring insights that make predictions and forecasts about the future.

For example, returning to the two charts in Figure A and Figure B, the question was what the next bar would be on the trend in Figure A: up or down? A viewer of the chart might lean to up because the general direction of the trend is up or due to a personal need/desire to have the trend continue up. However, applying the statistical process control index on the data in Figure A would predict the next bar to be materially down—which it was, as depicted in Figure B.

      This is a beautiful example of applied statistics to reveal an unbiased insight that can, and should, materially impact a decision. Whereas reporting and data visualization informs what happened and where it happened, analytics powerfully advises what will happen and how to make it happen. As we shall explore in depth in Chapter 5, using the full range of tools, decisions can be enhanced through information and insights that span a continuum of time in the past, present, and future.

BIASED VS. UNBIASED

      Most planning, budgeting, and forecasting are biased: that is, a value for the future that is based on a human's guess. While the guess may be from experience or gut feel, it is a value that is not mathematically calculated from past performance of the business. Biased forecasts are always fraught with human frailties because, as mentioned, they are about what we want or need the future to be. How many times have you made a spreadsheet and not liked the outcome displayed? Hardly ever, for most of us—we simply change the values and, voilà, get what we want. Biased decision-making will be explored further in Chapter 2.

      Many sales teams pronounce their “forecasts” with immense certitude by claiming the forecast is from the CRM system. The importance of the CRM is to establish the credentials of the source, like the Good Housekeeping seal of approval. It is authority, credibility, and accuracy all rolled into one. But—and this is a big but—the forecast is merely the sales rep's guess of when the deal will close.

      A company typically establishes a ranking system for where a sales deal is in the pipeline and its probability to close, but as disciplined as this ranking may be, it is not “analytics”—that is, it is not derived from the application of mathematics on data. The fact the sales rep enters the “forecast” into the CRM does not transform it to anything beyond a guess.

While sales reps are often good guessers, they achieve many of their forecasts, especially at the end of a quarter, through a modicum of “unnatural” acts that have deep discounts and concessions the business pays for in reduced profitability down the road.

      Analytics provides unbiased intelligence that is an essential input into decisions, as the mathematics of analytics is dispassionate. Formulas have no predisposition to a desired outcome. Data about the past is historical. As such, the combination of math and history yields a view to what the future can be vs. what one wants the future to be.

      Business needs human intuition, as we have a good sense of what is around us, but we are biased about what is ahead of us. As such, when looking forward, there is a fundamental need to incorporate unbiased predictions and forecasts that can be gained from analytics. When the two are combined, the man-and-machine efforts produce higher accuracy predictions over a longer time horizon.

AI AND ROI

Research typically pegs the ROI on analytics at a minimum of 10X. For example, according to Boston-based NUCLEUS Research, the 2014 survey on Analytics ROI revealed that the average return “has increased to $13.01 for every Analytics dollar invested.”⁵ An excerpt from a November 2011 Research Note from NUCLEUS Research highlights the visibility that analytics provides:⁶

      Software buyers may think that vendors overhype visibility as a benefit of analytics, but Nucleus found that, in fact, the highest-ROI analytics deployments made data more available to decision makers and enabled them to find ways to increase revenues or reduce costs. Nucleus found analytics enabled improved visibility in three areas:

       Revenues. The more managers knew about what customers where (sic) buying and why, the better able they were to accelerate sales cycles, cross sell, and maximize pricing.

       Gross margin. By serving up highly granular data on costs of goods sold, analytics applications helped decision makers identify the highest margin products so that they could push the right products and increase gross profit.

       Expenses. The more managers … learned [from] analytics … the better able they were to reduce or eliminate expenditures that were unnecessary or generated low returns.

As seen in Figure 1.3, the report “The Analytics Advantage, We're just getting started,” from Deloitte, reflects key findings from the Deloitte Analytics Advantage Survey, including “Nearly half of all respondents (49 percent) assert that the greatest benefit of using analytics is that it is a key factor in better decision-making capabilities.” Further, when asked “Does analytics improve competitive positioning?” some 55% of respondents indicated that analytics Fairly to Significantly improved positioning.⁷

Figure 1.3 Deloitte analytics for decision-making.

      With executives agreeing on the value of analytics for decisions and competitive capability, we note that business performance betterment projects must be measurable, and AI is no exception. To this end,