Evaluation and selection of software based on best-of-breed criteria has been given a lot of hype and is touted as a best practice; but is it really?

Best-of-breed is supposedly the best product in a given category of software, but by whose standards? The popular belief is that if we buy the best- of-breed product in each category, then we have the best-of-all-worlds software environment.

Yet, best-of-breed classification generally only considers functionality. Software selection on the basis of functionality alone generally increases information quality problems while decreasing the ability to capture all required information. Rarely do functional evaluations provide a true picture of the risks and costs of interfacing the package into the enterprise information environments. An expensive best-of-breed may increase costs to interface into the existing information environments. It may be the riskiest, most complex and represent the highest total cost of ownership in the form of maintenance and information quality problem recovery.

Information Quality Issues

The more complex the set of interface programs or even middleware agents to keep data "in sync" across their myriad islands of information, the greater the potential for errors to be introduced.

The most significant issue in software package evaluation and selection and subsequent integration is the potential for failing to control information quality. Definition of a fact in a package is that of the software developers and may be different from your organization's definition. Data synchronization routines must know when a record in one file is to be propagated and maintained in sync with another file. Different data element lengths causes truncation errors and loss of portions of the data. Different data element types requires data transformation to a form equivalent in meaning, such as how nulls are represented. Different valid value sets requires data value transformation across different databases. Complexity of data synchronization mechanisms forces many not to attempt it, creating two problems: increased costs of "multiple data capture and maintenance," forcing information producers in different areas (in different islands of automation) to create and maintain the same or similar data about the same real-world objects; and, redundant data not synchronized will likely not be equivalent, decreasing quality while increasing costs. The bigger risk is that data used in decision making will not be easily available or will be incomplete, inaccurate or even inconsistent or contradictory.

From both an information quality perspective and a business performance perspective, software should be evaluated with a "best-fit" criterion. In other words, what software best meets the organization's total requirements –­ functional, information and technical?

The organization must evaluate software against its enterprise data model as the representation of its information requirements. Evaluate the package data structure against the enterprise conceptual data model; and evaluate package data element definition, length and valid values to its attribute definition, length and valid values. This will enable the organization to understand how well the package meets its knowledge requirements.

Furthermore, the organization can better calculate the total costs of ownership of one package compared to another. The best-of-breed package in one category may have far more significant costs than a less function-rich package that requires few transforming, programming or agent interfaces.

Organizations that subscribe to a purchase-rather-than- build philosophy can truly manage their information environments only by maintaining a robust enterprise data model that represents their enterprise information requirements and using it to evaluate packages and control data distribution.

If an organization implements an enterprise database environment by subject (business resource) classification, it can use that to manage and control data movement from package to package. By having an enterprise database(s), one has a mechanism to control data movement. One also has the ability to swap out software packages, eliminate them or reengineer legacy applications as is feasible. The enterprise database serves as a hub-and-spoke control. All information interchange would come from an originating software package to the hub. The hub distributes the data to any redundant databases requiring it. This approach, described in Improving Data Warehouse and Business Information Quality (pages 278-282), is being used by world-class organizations such as Aera Energy as a way to control information and quality in a multiple software package environment.

Best-fit, rather than best-of-breed, is a better approach to control IQ in the multiple software package environment because it puts the customer organization in control of its information requirements; maximizes control of data distribution; and provides an economical way forward to replace software packages, eliminate them or reengineer obsolete legacy systems.

What do you think? Let me know at Larry.English@infoimpact.com.

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