|
|
Intelligent Enterprise
August 1999
What's It Worth
Richard Winter
I usually examine issues of technology, architecture, or practice in this column — focusing especially on those problems that get much more challenging or interesting as scale increases. Occasionally, though, I like to look at the related business questions. And lately I’ve heard a lot of talk in the industry about the business value of data warehousing, something that I argue can be measured effectively, despite popular doubts.
Of course, the larger the data warehouse, the bigger the questions about its business value. You might even argue that this scalability problem is worse than linear. That is, a data warehouse that costs $10 million to implement may be more than 10 times as hard to justify than one that costs $1 million. Despite its importance, however, two factors seem to prevent careful consideration of a data warehouse’s business value: taking for granted that it’s a good investment, and thinking that the business value is prohibitively hard to define, anyway. The first mistake people make is misunderstanding the meaning of “business value.” When asked what the business value of their data warehouse is, people often respond with something like, “Better access to information.” But ask a business executive — perhaps a CEO or COO — what business value is, and you’ll generally get an answer in terms of financial performance.
If you want to talk about hard business value in the real world (leaving aside, for a moment, the world of Internet stock values), you generally need to talk about either profit or its direct components: revenue and cost. Usually, profit is the thing in which business executives are truly interested. The proverbial “bottom line” is at the core of hard business value in most enterprises.
Soft business value will buy you less than hard business value, but it’s better than nothing. Most executives define soft business value as positive changes in measures that are accepted in business — or in your business organization specifically — as strong influences on financial performance. Some examples of those measures are market share, customer satisfaction, delivery time, product quality, and inventory age.
If you’ve been classifying soft measures as hard, then consider a true story in which a bank defined a campaign’s goal in terms of the number of new credit card accounts they acquired. The new data warehouse brought the issuer a large number of new customers each month — but it was losing more money each month, as well. Why? Because most of the new customers didn’t purchase anything with the card. Many were keeping it for only a few months, so the company was acquiring customers at a loss and losing them before it could yield a profit.
In fact, most of the “results” that IT professionals talk about are not statements of business value. “Better access to information” is not business value. Who knows how it affects profit? Maybe the information is not needed. Maybe the information is helpful but not valuable. Maybe the company spends more money making the information available than is gained by having better access to it. “Increasing system usage” and the popular business mantra, “knowing our customers better,” are likewise not business value statements.
The Application View
The key to understanding data warehouse business value is to view the warehouse as a platform for implementing business solutions.
As much as possible, you view and organize usage into a portfolio of business solutions. You then measure the cost and the value of each business solution.
What is a business solution? It can be an off-the-shelf decision support application, such as Valex (a campaign management product available from Exchange Applications Inc., www.exapps. com. It can be a desktop application developed in house to retrieve data from the data warehouse and calculate customer profitability. It can be an organized set of canned queries and spreadsheets used to analyze the impact of industry-oriented sales programs.
You can craft a business solution using any of the tools and/or resources of a data warehouse environment. It can be implemented by the IT organization, a user organization, an integrator, or a vendor. The important thing is that it presents a solution to an identified business problem and results in a concrete business impact.
It is my experience that enterprises create most of their data warehouse usage in the form of business solutions—and create those solutions because of identified business problems. Sometimes this is seen by the IT department as undifferentiated ad hoc use. But most “ad hoc query,” in my experience, is also really application solutions. If a skilled user in marketing creates a set of 10 canned queries that the brand managers use to understand the performance of their promotions, that’s an application.
There are applications that drive data warehouse use, and there are fewer applications that drive data warehouse value:
The Data Warehouse Value Equation
Let WV = the business value of a data warehouse
Let AnV = the business value of the nth application of the data warehouse
Let QV = the business value of the ad hoc query usage of the data warehouse
Then the equation for value is: WV = A1V + A2V + AnV + QV
The actual value, WV, is only ever known approximately. It’s not even important to know it precisely. If you propose to spend a million dollars on a data warehouse, you hope it will generate tens of millions of dollars in business value over the next three to five years.
My advice is to forget calculating QV. Usually, most of the business value will come from applications that provide defined solutions to business problems. Just view QV as a bonus. If you have the feeling that ad hoc query usage at your site is generating extraordinary business value, then look closer: You will usually find one or more applications hiding there in the form of canned queries or reports.
With regard to AnV, the secret is usually to analyze business value for a few of the “big hitters,” the ones expected to provide a large, quantifiable business value. Campaign management is a great one for this: Marketing campaigns have implicit quantifiable costs and revenues. You can measure them before implementing the application, project post-implementation measurements, and then evaluate again after implementation.
Ordinarily, a few major applications dominate the picture in terms of quantifiable business value. Of course, there are many others that may be too small or difficult to measure. You don’t need to measure them. You only need a rough equation for value.
Answering the Big Complaints
The complaint that it is impossible to define the business value of a data warehouse comes mainly in three forms, each of which I counter using the preceding equation.
The data warehouse will be used in thousands of different ways — too many to make it practical to measure business results. You don’t have to measure all the different ways it is used, just measure a few big ones. In most companies, the majority of the business value will come from a few major data warehouse applications. In my experience, you can usually project enough value from three to 10 applications.
We can’t predict how it will be used. You can ignore the unpredicted use, at least for the purposes of projecting the value of the data warehouse. As unpredicted uses materialize, of course, you can measure them and factor them into the value equation. But the question at hand normally concerns what predictable benefits can we get from some very predictable expenses. So focus on the anticipated uses and treat anything else, like the value of ad hoc query, as a bonus.
The data warehouse provides one component of a business advance; we can’t measure how much it contributes. Answering this protestation involves another important concept: changing the business process.
Business Process Change
The truth is that business value doesn’t come from the data warehouse. It comes from business process changes that are enabled by the data warehouse. Suppose you install a campaign management system. Now you can target your promotions with more granularity. Instead of a one-percent response to your direct mail campaigns you can get a two-percent response.
Now you can sell a million dollars worth of cameras with a $50,000 mailing instead of a $100,000 mailing. That’s a contribution to profit — an example of hard business value. But what produced the contribution? It wasn’t entirely the warehouse, was it? Even if you throw in the cost of acquiring and implementing the decision-support package, that doesn’t capture it either.
What produced the profit contribution was a change in the marketing process. Instead of mailing to everyone on our customer list who had ever bought an optical product, the marketing department approached the promotion in a different way. They asked a different set of questions to select the target. Because they were mailing to fewer people, they were able to consider a better, more attention-getting mailer. Because they tracked the results of each promotion, customer by customer, they were able to tell who was responding to what. So a substantial process change was involved.
And who produced these business results? The marketing manager charged with implementing the new process, experimenting with the techniques (including the data warehouse), and producing something that worked.
So the business value came from the implementation of change within the marketing process. The implementation of a campaign-management package supported the process change. The data warehouse — and the data that had to be added to it, integrated, or upgraded in quality — supported the implementation of the campaign management package.
The business value — AnV for the campaign management application — is the value of the change in the marketing process. And that is the answer to the third complaint: You don’t measure the data warehouse alone—you measure it as a part of a business process change and an application implementation. This view is the appropriate way to look at costs as well.
Applications Are the Key
Experience tells me that, with regard to data warehousing in general, applications are the key. They reside at the core of most performance and scaling issues. They are principal to most aspects of technical planning and have a profound influence on architecture. And even though enterprise data warehouses are intended to serve all possible uses, most technical decisions should and do consider the character of major applications that must be implemented, making applications the key to defining and quantifying data warehouse value.
If you adopt the application view, you can quantify business value. By
doing so, you can build support for your data warehouse program within the
client organization. Remarkably, this step creates a positive cycle,
whereby clients who understand the business value of the warehouse become
its advocates in new projects and, as a result, help increase the vitality
and the value of the warehouse program over time.
Richard
Winter is a specialist in large data base technology and
implementation, and is president of Boston-based Winter Corp. You can
reach him via email at Richard.Winter@wintercorp.com
or by fax at 617-338-4499.