http://www.intelligententerprise.com/010507/feat1_1.jhtml

Information Impact

This special series explains how to measure and quantify the value of your company's critical information.

By Erik Thomsen

Executive Summary
Erik Thomsen

Information is a critical asset to an organization. But how do you value it? Without proper methods of valuation, you cannot rationally decide among alternative IT (and other information affecting) investment strategies. Traditional attempts to quantify the value of information have focused on correlating what is most readily observable: IT spending and organizational performance. This first in a series of articles will help you understand how information creates organizational value, and how to quantify that value, by showing you how to measure key aspects of the utilization of information at key junctures along the causal links defined by decisions.

The boom in data warehousing and decision support coincided with the technology-driven bull market of the last half of the 1990s. During those high-flying times, it was almost impossible not to make money. Information - and the technology for collecting, storing, and managing it - was a source of competitive advantage. Although many companies conducted studies in an attempt to measure the ROI of data warehousing or decision-support systems (DW-DSS) projects, the major force driving the market was the fear of being left behind. I use the terms DW and DSS interchangeably because these days, there is enough overlap in their meanings.

For now, the heady boom times are over. You can no longer buy a new server, mumble a few incantations, and expect to receive a flood of new orders, customers, or investments. In these bearish days, it is increasingly necessary to justify all initiatives, IT or other, in terms of their quantified impact on your organization's key performance indicators.

So, how do you quantify the impact of IT investments on organizational performance to a skeptical and rational boss? Can it even be done? Traditional attempts have focused on linking the two things that are most readily observable: IT spending and organizational performance. While it is certainly necessary to measure IT spending and organizational performance, that isn't the whole story.

Crossing Hume's Chasm

The reason why performance isn't the whole story might be called "Hume's Chasm" (see Figure 1). As the empirical philosopher David Hume would have pointed out, although you may prove that $2 million were spent on a DW-DSS project and $6 million in additional revenues were recorded, these numbers do not prove that the spending on the DW-DSS project caused the increase in revenues. For example, sales may have increased because of a rapidly expanding market and would have still done so without the spending on the DW-DSS.

The only way to cross Hume's Chasm without falling in is to trace the utilization paths of information from IT systems on one end to organizational performance on the other. Decisions make up the causal links along these paths (see Figure 2). Thus, the only way to prove that changes to your information systems caused some measurable change to your organization's critical metrics is through a twofold approach: Link changes in measurable attributes of the information supplied with changes in decisions made, and link changes in decisions made with changes in your organization's metrics.

You can represent the relationship between IT investment and organizational value with linked ratios, where the D symbol represents a change. (See Figure 3.)

To link IT with organizational value, CIOs and CFOs should be shaking hands over decisions. CIOs should be asking, "What decisions will managers or executives make with the information generated by IT?" CFOs should be asking, "What were the decisions whose consequences are measured by our performance metrics?"

In an attempt to build a linked causal chain across Hume's value chasm, the purpose of this series is to outline an approach that enables you to measure or infer the following information value metrics:

  • Source information quality
  • Decisionoriented information quality
  • Decision quality per decision
  • Organizational value per quality of decision
  • Changes in decision quality per changes in information quality
  • Changes in organizational value per changes in decision quality
  • Changes in information quality per changes in IT investment.
IT questions


1. What is the total cost of your DSS-DW systems on an annualized basis?

2. Who uses the DW or DSS?

3. What information is most used? Least used?

4. How do you measure information quality?

Given these core information value metrics you will be able to:
  • Identify your most important decisions
  • Look at the information quality for your most important decisions
  • Determine which decisions rely the most on which information systems
  • Determine which decisions rely the most on noncomputer information
  • Assign a monetary value to the improvement of information quality
  • Evaluate the relative merits of investing in different information systems.

The approach complements any management metric program currently in place, whether based on the balanced scorecard, value imperative, value code, activity-based management, or any other approach. You can also use the approach with any method for propagating estimated costs and benefits such as net present value, risk-adjusted return on capital, decision analysis, and real options (see Resources).

ORGANIZATIONAL METRIC QUESTIONS


1. What primary metrics does your organization - and its eternal analysts - use to measure its overall performance? Typical examples include sales, sales per employee, earnings before interest and taxes (EBIT), gross profit, gross margins, and earnings per share.

2. What are your key secondary metrics or key performance indicators? Secondary metrics aren't tied directly to an organization's financial performance. Instead, they reflect specific operational areas. Typical examples include customer satisfaction, product quality, and employee satisfaction.

3. Over what range of operational dimensions can you evaluate your organization's primary metrics? Can you evaluate earnings for each customer or per average customer; earnings per product or earnings per employee?

4. How would you compare the value of actions that have direct impact on the organization's primary metrics, such as sales-to-lead ratios, with actions that have direct impact on the organization's secondary metrics, such as improved manufacturing quality or upgraded retail space?

5. How would you compare the value of actions that affect only the organization's secondary metrics? For example, how would you compare the relative merits of improving customer satisfaction vs. product quality?

6. How would you rank your organization's metrics by degree of certainty? For example, cash on hand typically has a high degree of certainty, while customer lifetime value is composed of a variety of predictions and has a much lower degree of certainty.

Dss Self-Diagnostic Capabilities Quiz

Before you can improve the value of your DSS-DW systems, you need to be able to measure their current value. In order to measure their current value, you need to have the appropriate self-diagnostic capabilities.

The questions break down into organizational metric questions, IT questions, and decision questions (see the sidebars, "Organizational Metric Questions," "IT Questions," and "Decision Questions,"). By trying to answer the questions, you will discover your organization's DSS self-diagnostic capabilities. Note how all of the DSS value questions are in the decision category.

Congratulations if you can answer all of the questions! However, if you are like most readers, you would have a hard time answering most of the decision questions and even the more common organizational metric and IT questions. But if you can't answer all these questions, you do not have the ability to diagnose the state of your decision-support systems; you can't value the information that you are spending money to create, and you can't rationally choose between alternative information investments. Your organization's CIO and CFO are flying blind.

The DEER cycle Approach

The decision execution environment representation (DEER) cycle approach focuses on:

  • Decisions as the key information utilizing events
  • Decision consequences as what are (or should be) measurable by the organization's value metrics
  • The linkages between changes to:
    • Source-oriented information attributes (such as timeliness)
    • Decision-oriented information attributes (such as understood correlations between information values and optimal decisions)
    • Decision quality
    • Organizational values that could be financial or any other type of indicator.

    The DEER cycle approach provides for varying levels of precision. It works just as well for information that resides in people. And it doesn't assume that decisions are the only source of value creation for the organization, but rather that decisions are the only source of value creation for factual information (such as sales data), as opposed to skill-enhancing information (such as online tutorials). Finally, the DEER cycle approach introduces a framework for looking at decisions and highlights the fundamental importance of good dimensional analysis in order to evaluate organizational metrics over the greatest location range possible.

    The approach works by defining causal connections between changes, which affect measurable attributes of the information and changes to the measurable values of key organizational metrics. As shown in Figure 4, the DEER cycle connects measurable organizational metrics with the outcomes of key, activity-based, decisions within the framework of decision graphs that define the incremental organizational value of incremental changes in decision quality. Such changes in decision quality connect with incremental changes to the measurable attributes of information.

    I will describe the DEER cycle approach in the remainder of this series along with the following three supporting topics, which you will also need to understand:

    • The phases of a decision cycle
    • How you can evaluate causal linkages through multiple passes of increasing precision
    • The interplay of dimensional modeling and organizational metrics.

    As I would like you to get a quick feel for looking at decisions as the causal link between information and organizational value, I will end this first installment with a small example of how to use rough weighting, which is the coarsest form of valuation, to get a fast thumbnail estimate of the value of certain decisions, the value of the underlying information, and the quality of the information upon which you base your decisions.

    Decision questions


    1. How would you calculate the total value to your organization of your current DW or DSS systems?

    2. How would you chunk and order the data and processes in the DW or DSS from most to least valuable?

    3. How would you calculate the average information quality for your organization's most important decisions?

    4. What percentage of the value of your most important decisions is attributable to the DW or DSS? To answer this question, you must know the value of a decision and the value of the information that contributed to the decision.

    5. What are the key non-DW or DSS sources of decision value?

    6. Which decision improvements could provide the greatest additional value to your organization as a result of noncomputer-based information improvements?

    7. Which decision improvements could provide the greatest additional value to your organization as a result of improvements to your current decision-support infrastructure?

    8. What are those changes?

    9. How would you calculate the value of those changes?

    A ROUGH WEIGHTING EXAMPLE

    Imagine, for example, that you're trying to get a quick estimate of the relative importance of the information derived from focus groups for apparel purchasing vs. the information derived from short-term sales forecasts for in-season campaign management; and compare the quality of the information with the importance of the decisions.

    The first steps you need to take are:

    1. Define your primary organizational metrics.
    2. Weight whatever business activities are under analysis according to their impact on the metrics.
    3. Weight key decisions by their impact on the activities.
    4. Weight decision-making information you relied on by its impact on the decisions. Because this is a rough weighting, you should not try to be overly precise. Give yourself a realistic margin of error, such as +/- 25 percent, which means that you can only clearly distinguish about five different relative weights such as: 5, 10, 25, 50, and 75.

    Step 1

  • Assume the organizational metric is apparel earnings with a typical annual value of $10 million. And that for steps 2, 3, and 4, you've assessed the relative weights as follows (presumably through interviews with key domain experts):

    Step 2

  • Purchasing, in-season campaign management, and post-season campaign management as 70 percent, 25 percent, and 5 percent.

    Step 3

  • Within purchasing, the relative weights of the key decisions of what to purchase and who to purchase from are 75 percent and 25 percent respectively

  • Within in-season campaign management, the relative weights of the key decisions of when to hold promotions, what to promote, how much to discount, and how to promote are 50 percent, 30 percent, 10 percent, and 10 percent

    Step 4

  • For the purchasing decision, the key information is nine-month forecasts, shows, customer focus groups, and hunches with relative weightings of 50 percent, 20 percent, 20 percent, and 10 percent

  • For in-season promotion time decisions, the key information is sales to date, next month sales forecasts, competition, and the economy with relative weightings of 70 percent, 20 percent, 5 percent, and 5 percent.

    Step 5

    For every piece of information identified in step 4, estimate its coverage, accuracy, and timeliness on a scale of 0 to 1 and multiply the three attributes together.

    The value calculated in step 5 represents the source-oriented information quality shown in Tables 1 and 2. The values in the "Aggregate Quality" column are the product of the values in the "Coverage," "Accuracy," and "Timeliness" columns. The "aggregate weighted information quality per decision" value is the sum of the products of "aggregate quality" and information "weight."

    Step 6

    For every piece of information, estimate the understood correlations relative to the decisions that you need to make. This estimate is the decision-oriented information quality and is not shown in this example.

    Information Valuation

    Multiplying any of the decision weights listed in step 3 with the activity weights listed in step 2 and the organizational value listed in step 1 yields a rough value for the decision. Thus, for example, the rough value of decisions about what who to purchase from are $10,000,000 * 0.7 * 0.25 = about $1,750,000. By rough value, I simply mean the organizational value connected to the decision. On average, the greater the rough value of the decision, the more you have to gain or lose by improvements or degradations to the decision.

    Multiplying the information weights by the cumulative decision weights yields the rough value of the information. For example, the rough value of "customer focus group" information as used for decisions about what to purchase within purchasing activity is $10,000,000 * 0.7 * 0.75 * 0.2 = about $1,000,000, while the rough value of "sales to date" for in-season promotion campaign timing decisions is $10,000,000 * 0.2 *0.5 * 0.5 = about $500,000. So, even though sales-to-date information is more than twice as important for making decisions about when to put on in-season promotions than customer focus group information is to making decisions about what to purchase, purchasing decisions are so important to overall earnings that focus group information is twice as important as sales-to-date information.

    Comparing the most important decisions by weight with the weighted average quality of the information they rely on lets you focus your troubleshooting energies on those important decisions with the poorest average quality of information. For example, notice how the aggregate quality of purchase decision information as calculated in Table 1 is about half as good as the aggregate quality of campaign decision information as calculated in Table 2, while the purchasing decisions are about twice as important as the in-season campaign timing decisions. Someone should address the imbalance.

    Although the valuations are very rough, they only assign relative weights, and they do not estimate the value of incremental improvements or degradations to your information. The process of defining relative weights forces you to figure out how things connect to each other, (in other words, your organization's decision topology), the direction of impact, and the relative impact size.

    This information is a critical first step in assessing the value of your organization's information and the decision activities in which it is used. And with this information you can identify high-impact areas where you need to have a more fine-grained understanding of information valuations.

    In the next installment, I drill into the DEER cycle, explore its phases, describe the process of decision reconstruction, and lay out a more precise way to estimate the value of information.



    Erik Thomsen, [erik@dimsys.com], was cofounder of Power Thinking Tools, which developed the first OLAP engine with integrated statistics, visualization, text processing, and object management. He is a researcher and consultant for Dimensional Systems and focuses on integrated multitechnology analytic solutions. He is author of OLAP Solutions (John Wiley & Sons, 1997) and coauthor of Microsoft OLAP Solutions (John Wiley & Sons, 1999).


    RESOURCES

    Boulton, Richard E.S., Barry D. Libert, and Steve M. Samek. Cracking the Value Code: How Successful Businesses Are Creating Wealth in the New Economy (HarperCollins, 2000).

    Dragoo, Bob. Real Time Profit Management: Making Your Bottom Line a Sure Thing (John Wiley & Sons, 1995).

    Kaplan, Robert S. and David P. Norton. Translating Strategy Into Action: The Balanced Scorecard (President and Fellows of Harvard College, 1996). Kontes, Peter W., Michael C. Mankins, and James M. McTaggert. The Value Imperative (Simon & Schuster, 1994).

    Violino, Bob. "ROI Intangible Benefits," Information Week, June 30, 1997

    Thoreson, Gregory D., Jeffry Thoreson, and J. D. Thoreson, "Economic Worth of Systems"


    Real Options Web Site
    Decision Analysis Society


  • Return to Article