How to Achieve Actionable BI
What to do with what we know is the most important question challenging business intelligence (BI) and data warehousing (DW) today. The extent to which corporations and governments are able to do something with the analytics from the warehouse will determine the future of BI/DW. It is at the heart of achieving actionable BI.
At the TDWI session on BI Strategies in November 2003, Deb Masdea, director of business information and analysis at the Scotts Company, used this question to highlight her BI solutions for fact-based decision-makers. Since then, that question has motivated me to probe into the issues of actionable BI.
In a past column that appeared in the November 2001 issue of DM Review, I discussed a framework for "complete" BI that suggested five maturing stages (observe, understand, predict, react and reorganize), as shown in Figure 1.
Figure 1: Stages of Complete BI
As illustrated by the line in the figure, there is a gap between the "knowing" and "doing" stages - a glass ceiling that prevents companies from maturing to actionable BI. What are the factors causing this glass ceiling? Let's explore a few.
Fire the Chauffeur
Analysts usually know what is happening in the business. BI tools and the data warehouse have been successful in providing insightful information about the business. The challenge is getting the right people to do something about it. Analysts do not "do" anything about what is happening in the business; they just analyze! They do not have the responsibility and authority to make decisions, kick some butts and solve the real problems.
Traditionally, there is a separation of function between the IT staff and line managers. We have moved analysts into the business units, but do we compensate them based on their performance relative to production goals?
Principle: Fire the chauffeur! The people that make your business happen need to drive the car (BI systems) themselves. The theme "BI for the masses" has some truth within it. BI tools should be designed so that anyone can use them and effectively analyze enterprise data. However, not everyone needs to know everything. Analytics must be tailored specifically to the job responsibilities in the language and culture of that job.
Squash the Pyramid
We need to kill a framework that, since the early days, has helped us understand (and design) enterprise systems but has grown too simplistic. It is the management pyramid, as shown in Figure 2.
Figure 2: Management Pyramid
The framework states that three levels - strategic, tactical and operational - neatly divide the responsibilities for managing an enterprise. Top managers make strategic decisions that generate policies. Middle managers make tactical decisions that generate procedures. Line managers make operational decisions that execute those procedures. For each level of management, the information requirements are distinct in terms of usage frequency, granularity, accuracy, time horizon and so on. For instance, top managers only need to see the big picture spanning months and years, and line managers should only care about production quota for this week.
However, because business processes are loosely coupled and globally distributed, exceptions dominate many of these processes. The best design of corporate policies and procedures often seems inadequate for the complexities of our businesses.
Principle: Squash the pyramid! The person responsible for a business process should view it strategically, tactically and operationally every moment of every day. Through BI systems, that person should have access to the whole context and nothing but that content. Regardless of the time horizon, granularity and the like, that person should understand every facet in order to make the best decision.
Minimize the Action Distance
For the people who touch customers, products and other key ingredients of the business on a daily basis, the action distance (AD) is zero.1 These people are in a position to initiate actions that can change important things. A critical issue is whether they have the skills, support, responsibility and authority to take that action. Although most issues are organizational and staffing-related, BI/DW can have a major impact. Better analytics can reduce the required skill level, and better analytics can add clarity to the situation.
Principle: Minimize the action distance! Make the distance between "knowing" and "doing" as close to zero as possible. In other words, the people who know should also do. The zero-AD people are your front line; allow them to sparkle.
Leverage the Multiplier
We need to intelligently set our priorities. What is really important, and what is not? The answers will determine where to focus BI/DW support.
As BI professionals, we tend to overstress the big decisions, which we glorify as being strategic. Yes, there is a lot of money behind big decisions. However, they are infrequent and also inefficient in contributing to the bottom line. We take too long to make big decisions and then spend an equally long time evaluating and reevaluating the decisions.
Instead, we need to save a nickel by focusing on each instance of a key business process, thousands of times a day in hundreds of locations. If we use analytics to honestly determine where the real impacts to the business can be made, we will probably be directed to the lowly, frequent little decisions. Every day when assigning crews to their 153,272 freight cars, Union Pacific calculates the total value of goods (along with penalties for missing shipping deadlines) on thousands of their trains. A nickel here; a nickel there.
Principle: Leverage the multiplier! The multiplier is closer to the money flow and provides an easier way to make a real impact. The action is not in the fancy headquarters building, but in the dirty shipping yard, so wear your boots.
Alert, Inform and Guide
We design our analytics incorrectly! The goal is not to deliver information. The goal should be to support decision making that results in action. I suggest that this support must: alert, inform and guide (AIG). Alerting implies that a person's normal workflow should be interrupted to focus on a more important situation. Informing implies that the entire context is distilled so that the person can quickly understand the situation. Guiding implies that courses of action are outlined along with the means to quickly affect those actions.
How To Achieve Actionable BI
Principle: AIG it! Provide the functions of alerting, informing and guiding as part of the infrastructure for delivering any report or analytic. Achieving this will require a huge commitment to the BI/DW infrastructure.
This column has explored the barriers to achieving actionable BI. Five principles were offered that might reduce the barriers. Some of the issues are technology-based, but the majority will involve a deep rethinking of organizational design and management styles. Therefore, the continuing success of BI/DW depends on the willingness to move out of our corporate technology comfort zone and into the agony zone of corporate politics. Masdea remarked, "Companies will be more successful with BI if they find the right persons who can cut through the politics."
We will stay on the right track if we keep asking, "What do we do with what we know?"
1. For more on Action Distance, see R.D. Hackathorn, Minimizing Action Distance. DM Review September 2002. http://www.dmreview.com/article_sub.cfm?articleId=5679.
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access