Today's highly competitive business world is becoming increasingly complex, and the dramatic growth in the use of the Internet as a sales channel means that business managers need to be able to react in Web time to changing business conditions if they are to remain competitive and meet customer needs. In this volatile business environment, corporate decision-makers need rapid and easy access to in-depth analysis of business operations if they are to make informed business decisions. To satisfy this unending demand for information, IT groups have rushed to implement data warehousing systems, but there have been as many failures as successes.

Key Issues

Although there is no question that data warehouse technology and business intelligence tools have been successful in helping many organizations reduce costs and increase revenues, there are nevertheless several key issues that need to be addressed by current data warehousing solutions if they are to evolve to support the new world of e-business. The three main ones that vendors need to address are:

  1. Project implementation time and cost.
  2. Business intelligence tool complexity and user difficultly in accessing business information.
  3. Poor data quality and incomplete business information.

In this article, we take a detailed look at these three issues and show how the use of packaged analytic applications can help resolve them and thus improve the return on investment (ROI) of data warehouse and business intelligence projects. We also explain why it is necessary to be careful in selecting the right application package and offer some suggestions of things to look for in packaged solutions for corporate decision making.

Reducing Data Warehouse Development Time and Costs

Few things in life come for free, and a high-quality data warehouse is no exception. Extracting, cleaning and integrating information into a data warehouse from the many disparate data sources that exist in an organization is no mean feat, and this is why data warehouses are often expensive and time-consuming to implement. To reduce design and development time (and thus costs) many organizations are beginning to employ off-the-shelf packaged solutions for both building the data warehouse and also for analyzing and reporting on business information managed by a data warehouse. This trend will accelerate as the use of e-business applications increases and organizations discover that the build-your-own approach to data warehousing and business intelligence is too time-consuming and resource-intensive to satisfy the decision processing requirements for supporting e-business operations.

The two main arguments often used against packaged solutions are: 1) that the organization's information requirements are so unique that a packaged approach cannot satisfy those requirements, and 2) that packaged solutions create independent islands of information that are difficult to integrate into the organization's decision processing environment.

The Information Requirements Issue

The information requirements of an organization are not always as unique as they may first appear. Many business issues are common across companies. While it is unlikely that a packaged approach can satisfy all of the information needs for a specific business area in any given organization, it can nevertheless provide a high percentage of that information. The key to success in deploying packaged solutions is to select those products that satisfy 80 percent of the requirements and can easily be enhanced and customized to support the missing 20 percent. Where information needs are unique to a specific industry, vertical industry packaged solutions that are based on the best business practices of that industry should be employed.

The Information Integration Issue

While experience shows that packaged applications encourage more focused projects that are faster and cheaper to develop, their use can lead to data integration problems. These problems are caused by each package employing its own independent business and data models and data warehouse information stores (often called data marts). The solution is to build a shared information staging area that is used to feed extracted source data to underlying applications packages (and their data mart information stores) and to base the staging area design on a common and evolving business model of an organization's information requirements. Packaged analytic application solutions should be evaluated with this approach in mind to ensure that the package can be integrated into the staging area architecture and can be customized to support the common business model of the organization.

Making it Easier for Business Users

The traditional approach to decision processing is to build a data warehouse and then supply business users with a set of business intelligence tools for accessing the information in the warehouse. This approach may be acceptable for basic query and reporting or for experienced users, but it does not work for business managers who need detailed analyses and who do not have the time or experience to master complex business intelligence tools. Even when business users feel comfortable with a particular tool, they often find it difficult and time-consuming to navigate a data warehouse to find the information they need.

Canned queries and reports can help reduce the learning curve for using a business intelligence tool, but this approach is only a partial solution for business users who need to do in-depth analysis involving drill-down queries and who need to model different business scenarios. A better approach is to employ turnkey and Web-based analytic application packages that are designed to provide comprehensive analyses for the business area being researched and which offer a familiar and simple Web interface for the business user.

Satisfying Business-User Information Requirements

Many data warehouse systems simply don't provide the information required by business users. Data warehouses frequently are constructed based on the information that currently exists in transaction systems and with limited input from the business community regarding information requirements. This approach may be satisfactory for operational reporting but is inadequate for providing a comprehensive analysis of business operations for corporate decision making. The issue is that during data warehouse design, business users cannot always express in data terms the information they need. This is where the best business practices built into analytic application packages can help. The predefined business metrics supplied with analytic packages can be employed in brainstorming sessions between data warehouse developers and business users to fine-tune information requirements.

In many organizations there are a handful of key business metrics (for example, revenue dollars per sales rep per day) that are employed by business managers to monitor the health of the company or to determine the success or failure of sales campaigns, new product introductions, and so forth. The business rules behind these metrics are often very complex, and it is often difficult to configure business intelligence tools to provide these metrics. Analytic applications designed to support corporate decision making can ease the burden of building applications to create and maintain these metrics. A key requirement is that an analytic application should store its metrics, and the business rules behind those metrics, in an open repository so they can be customized to suit the organization's information and business requirements as they change and evolve over time.

Analytic application packages, however, are not all created equal. Many of them provide only a superficial set of business metrics that offer little more than can be easily produced by a standard business intelligence tool. This is why many IT groups often oppose the use of packaged solutions ­ it is often quicker to build the application in house. The bottom line is some packages are designed for basic workgroup or line-of-business reporting, while others are intended for corporate-wide decision making. This must be taken into account in the selection process (see Figure 1).

Figure 1: Types of Analytic Application Packages

Another problem facing the developers of decision processing applications is that the required information often does not exist in the transaction source systems but is instead stored in collaborative documents such as spreadsheets, word processing documents, and so forth. This will be even more true as corporations deploy closed-loop decision-making systems where the decisions made based on the business metrics supplied by analytic applications are fed back to transactional and e-business systems via collaborative processing documents (see Figure 2). Another growing trend is the use of information provided by external providers such as Axiom and ACNielsen. It is important, therefore, that analytic application packages can support sources of data other than just from transaction processing systems. The use of staging areas also helps since nontransactional data can be fed into the decision processing system via these staging areas.

Figure 2: A Closed-Loop Decision Processing System

Choosing the Right Analytic Package

So far we have addressed some of the key issues facing data warehouse developers today and reviewed the benefits of using packaged analytic applications. We now move on to summarize the main selection criteria for evaluating vendor offerings in this area.

  1. What business areas and industry sectors do the application package support? All analytic packages focus on a specific business area (sales analysis, revenue management). Some packages support cross-industry applications, while others focus on a specific industry sector (banking or high tech). In general, packages for a specific industry sector are more likely to provide the detailed analyses required by corporate decision-makers.
  2. Is the package part of an integrated application suite? In the transaction processing environment, integrated suites of application packages have dominated the marketplace. In the decision processing environment where the range of requirements is very broad, it will more difficult for a single vendor to provide a suite of analytic applications that completely satisfy an organization's needs. It is very important, therefore, that the package support open interfaces that allow its integration into an organization's decision processing environment.
  3. Is the package designed for the power user or the executive? Some packages are designed for business analysts and power users, while others are designed for the executive. In both cases, the package should be Web-based and capable of being personalized to suit the needs of the business user and his/her job function.
  4. What types of analytic capabilities are provided with the package? The package should contain predefined business metrics (and underlying business rules) that can be customized and enhanced to suit the needs of the user (the user may want to add to the package, for example, business rules that are currently embedded in source documents such as spreadsheets). Customization also permits the package to be updated as information and business requirements change and evolve over time. The package should also permit analyses that allow the user to drill down from summarized to detailed information and to model different business scenarios.
  5. Does the package provide customizable data warehouse templates to help with data warehouse design and source data extraction and transformation? The package should provide a preconfigured data warehouse that is populated via a staging area of extracted source data. The package's data extract templates and data warehouse data model should be customizable.
  6. What data sources do the extract templates of the analytic application package support? The package should support extraction from multiple data sources including transaction processing, e-business and collaborative systems. Where a wide range of potential data sources exists, the package should support a standard input file format that can be used to load source data in a staging area for subsequent population of the underlying data warehouse.
  7. What interfaces does the package provide for information and meta data interchange with other products? As previously mentioned, it is highly unlikely that a single vendor can provide a complete decision processing solution for corporate decision making. It is, therefore, essential that an application package provide open interfaces that support the interchange of business information (and the meta data that describes it) with other decision processing products.

Vendors are now beginning to supply powerful packaged analytic solutions that reduce IT costs and development resources compared with in-house solutions. These application packages have significant potential for improving corporate revenues and profits. Organizations, however, need to select products carefully to ensure that a package can provide the comprehensive metrics needed to satisfy business-user requirements and can be customized and integrated into a cohesive decision processing environment.

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

Don't have an account? Register for Free Unlimited Access