Business intelligence dashboards

There are two dimensions to a dashboard: focus and interactivity. Focus is meant to summarize internal operations or external market forces. Interactivity refers to the ability of the business manager to do something with the dashboard s contents, the display itself. A static dashboard allows no interactivity. A business manager has to go elsewhere in the organization to learn more about what is displayed. A dynamic dashboard is fully interactive allowing that manager to learn more through point- and-click and drill-down. Drill-down is the process of disaggregating data, going from one level of detail in a data structure to a higher level of detail. For example, it is the process of going from sales in a marketing region to the individual cities that comprise the region. The regions are the low level of detail and the cities are the high level of detail. Both dimensions are discussed in the next subsections. See Lemahieu et al. [2018] for an explanation of drill-downs as well as roll-up, slicing, and dicing of a data set.

Dashboard focus

Dashboards have been a common part of the business environment for some time, especially since the penetration of IT into business. They are supposed to show the current state of a business so that decision makers can understand the state of their business and have insight into where they may have to take action. There are two types of dashboards: operations performance and market performance. The former is concerned with the way the different operations of the business are currently performing and perhaps how that performance has been changing over time. The performance is that of the different parts of the business. Recognizing that any business is not just a “business” but the sum of many interacting parts, it becomes clear that each of those parts must function well and in concert with other parts. The trite saying “like a well-oiled machine” is applicable here. If one part malfunctions, is not meeting targets, or is inefficient, then other parts dependent on it will also malfunction and the business could then be in jeopardy. A performance dashboard tells management how all the parts are functioning. This type of dashboard is internally focused.

A market performance dashboard is externally focused. It is concerned with how the business is performing in the market: how much is being sold; how much is being returned and why; how revenue, both gross and net, is changing over time; how the business’s product is selling by marketing region and customer classes; and the list goes on. Market insight gained by these dashboards is as important as that gained about the internal operations of the business. A business could function internally as a “well-oiled machine” but be losing sales or market share and not meeting sales and financial objectives. Its market performance will impact shareholder equity which could affect its ability to raise capital to further grow the business and, which is the focus of this book, innovate and develop new products. Management needs this type of dashboard as much as it needs an operations one. Since this book is not concerned with business operations, I am only concerned with the market-oriented dashboard and its follow-up issues and implications. I will classify the market-oriented dashboard as providing Business Intelligence, although it has to be stressed that the operations dashboard also provides Business Intelligence. See Eckerson |2010] for a thorough discussion of dashboards.

Dashboard interactivity

Whether internally or externally focused, a dashboard supplies information. Where this information lies on the Poor-Rich Continuum depends on the dashboard construction. See Few [2006] for the effective design of dashboards. The level of the information, how deep managers are allowed to go into the data, and the types of analysis they can do is another issue addressed by levels of interactivity. At one extreme, there is no interactivity so what managers see is what they get. The dashboard is static, to be updated only when the dashboard provider (i.e., the IT department) repopulates its data fields and regenerates tables and graphs. There may be a daily schedule for doing this, but nonetheless it is not the managers viewing the dashboard who do it. Any question or concern they have at the moment of viewing the dashboard will have to be addressed by someone else at a later time. This delay is not without a cost. If the new product is not performing well as indicated by the static dashboard, then any delay in learning why may result in the product failing. In short, time is of the essence.

A fully interactive dashboard allows managers to click on an image on the display and drill down for more penetrating insight; i.e., Rich Information. They could, for example, click on a bar of a bar chart to see further displays and reports on the data behind the bar. This allows them to learn more and make better use of their time to gain information. There is, of course, a limit to how far they can drill down, but however far they can go is better than what is possible with a static dashboard. The highly interactive dashboard is dynamic, changing at the control of the user.3

The limits of business intelligence dashboards

Both types of dashboards, operations and market, static and dynamic, provide business intelligence. They tell management what did happen or what is currently happening. These are important, but they beg the question management would eventually ask: “ What do we do with this intelligence?” In short, they need to know two more pieces of information:

  • 1. the root cause of any problem revealed by the dashboard; and
  • 2. what can be done to correct the problem.

The first requires a drill-down to reveal further insight which could be provided by a dynamic dashboard. Even with this type of dashboard, however, deeper analysis may still be needed beyond what managers are able to or should do. Everyone has two personal restrictions. The first is his or her knowledge of advanced statistical, econometric, data visualization, and machine learning methods to be able to go further. After all, people do specialize in areas and a business manager’s specialization is in running the business, not doing analytical work. At the same time, a manager, like everyone else, has a time constraint: you have only 24 hours in a day regardless of who you are. If you allocate your time to doing analytical work, then that is time taken from running the business; there is a time misallocation.

The second piece of needed information, the solution to the problem, requires predictive modeling allowing what-if or scenario analysis. This is also a specialty, probably more so than many forms of data analysis most managers think about. Some examples will be described below.

The following sections will illustrate some deep drill-down capabilities and predictive modeling that could be done. This only skims the surface, but it at least illustrates possibilities. This will be done using a case study.

Case study

A leading household furniture manufacturer sells to locally owned, boutique retailers throughout the U.S. which is divided into four marketing regions consistent with U.S. Census regions. The company has 43 products in six product lines consistent with the major rooms in a house: Den, Dining Room, Kids’ Room, Kitchen, Living Room, Master Bedroom. Each product line is divided into a product class such as Chairs, Tables, and Baker’s Racks for the Kitchen product line. Four types of discounts are offered at the discretion of sales force: Order Discount, Special Competitive Discount, Dealer Discount, and a Pickup Discount.

The product manager for living rooms has a problem with a new product: living room blinds (a.k.a., window treatments) with a remote voice control system that runs through a controller such as Alexa. Basically, the window blinds will open and close or rise and fall by voice commands. The product is one of the first in its class so there are few competitors. In terms of the Chapter 6 classification of new products I used, this is a not-new-to-the-world (NNTW) product.

The product manager keeps track of the product’s performance via a dashboard; this is Business Intelligence. The dashboard has displays of unit sales, gross revenue, returns, and net revenue (revenue based on unit sales less returns). These metrics are shown as bar charts with metric targets overlaid on the charts to indicate if targets are met or not. She has noticed that sales and revenue are not performing as expected so she has asked the Data Science group to investigate. Specifically, she needs information on:

  • 1. Sales patterns by:
    • • Marketing Region
    • • Customer Loyalty
    • • Buyer Rating: basically, a good or poor customer.
  • 2. An estimate of price elasticity based on sales as opposed to what she has from the pre-launch studies.
  • 3. Tests for statistical differences among the four discounts that are offered.

She has also requested a tool for predicting sales to a specific customer based on the customer’s characteristics. She needs this information so she can decide what to do to fix what she perceives is a marketing problem, not a product design problem.

< Prev   CONTENTS   Source   Next >