|
Data Quality |
|
Source: IT Business Edge | Priority:
Integrating the Enterprise |
Topic: Data Quality
Date Published:
9/22/2005
|
With William McKnight, senior vice president of Data Warehousing Conversion Services International and a frequent best practices judge for organizations involved in business intelligence and data warehousing, such as DMReview, TDWI, and the Software & Information Industry
Question: Data quality is surely one of the most important factors in any successful business intelligence initiative. What are the top best practices when it comes to data quality?
McKnight: Number one is having a data quality program. You need a means by which you can evaluate the quality of the data based upon corporate rules that are gathered from the respective business areas and that account not only for the common deviations from quality, but also the specific deviations from quality that would have a negative business impact on that company.
Number two would be business involvement in the data quality program. This is not an IT endeavor. The business is ultimately responsible for the quality of the data. I believe that IT needs to guide the program and lead the program, but the rules come from the business.
Question: Some businesses have half a dozen names for the same part. How can a business evaluate when to put in the manual effort to fix situations like that?
McKnight: The general rule is that each business should have one name and one code set for a given distinct business element. Often times we think we have different names for the same thing but we find that we have different business elements involved. In that case, it's perfectly proper to store each element discretely. For example, a company might have different color palettes for its products depending on whether it's marketing or manufacturing that needs the information. It's not that the company is confused about which color palette it's going to follow, it's that the company really does have two different color palettes and they both should be stored. When a business has these types of close calls, it should store both sets of data, give them proper names, and identify those names to the business through metadata, description and training, so that the business does not become confused.
Question: What are the most promising developments in data quality automation? What can computers do now that only human beings could do a few years ago?
McKnight: I think one of the most promising developments is the consolidation that we're seeing right now in the data quality marketplace. I think we're seeing data quality vendors becoming part of end-to-end vendor alliances because it's now considered part of the mainstream of business intelligence. What these tools bring is the ability to automatically cleanse data according to the rules of the business. What it does not do is absolve a company from setting the rules and having its business units involved in data quality.
|
 |
|
|
|
Hosted CRM Comparison Guide
This informative hosted CRM guide provides points of comparison on the solutions from these four vendors, which include pricing per user, service and support, and marketing features.
|
|
Selecting the Right BI Vendor
TAKEAWAY: The business intelligence market saw a barrage of consolidation last year, leaving four major contenders to the BI throne. This piece sets out to help companies choose among the big four and find the right solution to fit their BI needs.
Source: InformationWeek |
Priority: Aligning IT & Business Goals |
Topic: Business Intelligence
Date Published: 4/12/2008 |
Date Reviewed: 4/28/2008
'Single Version of the Truth' Doctrine Questioned
TAKEAWAY: This article deals with what happens when the idealistic quest for an enterprise-wide single version of the truth comes up against the realities of corporate life. "Once the major business actors realize what is at stake" (presumably raises and bonuses), data administrators have little hope of success. One tactic is to come up with broad data definitions everyone can agree on, but these are usually too general to have practical use. The better option is to accept that different business contexts demand different versions of the truth.
Source: DMReview |
Priority: Integrating the Enterprise |
Topic: Data Quality
Date Published: 12/1/2006 |
Date Reviewed: 12/8/2006
Metadata Is Your Friend? Maybe
TAKEAWAY: "Knowledge workers, metadata is your friend." That's the theme of this article, but it's a tough sell. It may well be true that individuals spend an average of 2.5 hours per day hunting for "information nuggets," but getting these people to fill in the metadata on enterprise content management (ECM) screens — or see the connection between that process and saving time — is another story. Still, there are lots of useful tips in this article. Example: When creating metadata categories, think about how you yourself would look for something.
Source: DMReview |
Priority: Integrating the Enterprise |
Topic: Structured Data
Date Published: 10/6/2006 |
Date Reviewed: 10/16/2006
|
|
|
|