-
Marketplace
-
Channel Resources
Articles from this Site
Talend Announces First Open Source Data Profiler
Equinox Pumps Up Data Quality with DataFlux
Departure Selects Alterian
Apatar Introduces CDYNE Death Index Data Quality Service
DataFlux Releases Version 8.1 of its Data Quality Integration Platform
White Papers
Data Warehousing Ensuring Data Integrity
Making Data Work: Addressing Data Quality at the Enterprise Level
Can your SharePoint Backup Harm Your Business?
The Value Behind Integrity
Building Profitable Customer Relationships and Personalized Retention Strategies
Web Seminars
Books
Corporate Information Factory, 2nd Edition
The Data Warehouse Challenge: Taming Data Chaos
Data Quality for the Information Age
Improving Data Warehouse and Business Information Quality: Methods for Reducing Costs and Increasing Profits
Metadata Management for Information Control and Business Success
Software Packages and Information Quality
Plain English about Information Quality
Amazingly, application software packages continue to be a "standard approach" in many organizations for delivery of information system solutions. Ian Sinclair wrote the following reply to my October, 2003 column, "How Best-of-Breed Software Selection Causes IQ Problems," while he was with the UK Ministry of Defence [British spelling] (MoD). I have only added clarifying wording with information quality (IQ) observations at the end.
I have spent the last two years trying to convince the UK MoD of the hidden IQ problems that are part and parcel of best-of-breed software applications. Unfortunately, my senior management sees these applications as "silver bullets" and cannot see beyond the functionality offered by the vendors. In addition to the points you raise in your article, in my experience, a major area that is overlooked is information presentation quality and the associated training costs.
The UK MoD has just written off £140 million (U.S. $255 million) spent unsuccessfully trying to implement an ERP (enterprise resource planning) software package for inventory management. [Note 1] This particular ERP package was developed in the U.S. for the nuclear power industry and had never before been deployed in a military environment. The first problems revolved around the business requirement to implement the vanilla product (i.e., no customization was allowed). This immediately set a challenge, as the language used in military logistics does not translate into the nuclear power industry. Therefore, knowledge-workers were confronted with business terms they did not understand. If this was not difficult enough, the application displayed different abbreviations of the same word on a single screen! [Note 2]
The application GUI was littered with information presentation quality problems. Field windows were far bigger than the maximum number of characters it was possible to enter. Therefore, knowledge-workers were led to believe that they had to enter more data than was actually required. My favorite problem concerned how text being entered would word wrap. One of the key data fields in inventory management is the Full Item Description field. In this particular ERP, two lines of text could be entered, but the first line would wrap after approximately 75 percent of the field window had been completed. Incredibly, different screens that displayed this data had to wrap in different places. [Note 3]
For the knowledge-worker, this was a nightmare. Imagine the potential problems if only a few characters wrapped onto the next line and this was not noticed by the knowledge-worker. Worse still, on some screens, only one line of the Full Item Description was displayed, so the potential existed that the knowledge-worker would not even be aware of any missing data. [Note4] Is this critical? In military logistics it is, as all our clothing is ranged. That is, the Item Numbers used to identify clothing run sequentially, incremented by one for each clothing size. For example:
- 99-123-1231: "Boots, Combat, Size 8"
- 99-123-1232: "Boots, Combat, Size 8.5"
- 99-123-1233: "Boots, Combat, Size 9"
The actual Full Item Description is far longer that the examples I have given, and the critical differentiator, the size, is usually at the end of the field. Therefore, if the knowledge-worker were to accidentally transpose the digits in the identifying number, he/she would call up a completely different item of clothing, but would be completely unaware as the application did not display all the data they required. [Note 5] I could go on.
I had the "pleasure" of visiting the vendors UK headquarters and raised all the IQ problems I had found. No prizes for guessing their response: "Well that's just how our developers have designed the screens." Their solution? "You'll just have to train your people." [Note 6]
And this is where the significant hidden costs come in. Introducing knowledge-workers to a new way of doing business is a significant challenge in itself, requiring significant investment in training. However, this is exacerbated when you have to teach knowledge-workers how to avoid the IQ pitfalls in an application purely because the vendor cannot be bothered to put things right. Then there are the unavoidable scrap and rework costs that would be generated as knowledge-workers struggled to become accustomed to the new software. Thankfully, this ERP was never implemented. [Note 1]
Information Quality Observations
- Note 1: Poor quality information design always results in costs, direct and opportunity - $255 million in direct costs, and lost time to implement an effective reengineering of important business needs in opportunity costs.
- Note 2: This reflects an absence of information standards. Data names and abbreviations must be intuitive to the business knowledge-workers.
- Note 3: The inconsistency of length of data fields on the screen versus on the database is a type of defect that would warrant a recall if it were in an automobile. This defect guarantees nonquality information.
- Note 4: This inconsistent information presentation, again, is a recallable design defect.
- Note 5: A way to minimize keying transpositions is to use check digits in primary identifiers.1 This prevents 93 to 94 percent of typical transposition errors from retrieving an incorrect record.
- Note 6: This is the antithesis of a "customer care" mind-set. I bet that this software firm truly considers the paying customers as "users" rather than valued customers and knowledge-workers.
You must evaluate software packages against your information standards and against your enterprise data model as a matter of due diligence for the sake of your knowledge-workers and the health of your company. You must find out these kinds of problems before your organization signs a contract for software. Remember, you are the customer, and you should expect the same warranties of freedom from harmful defects as if you were purchasing an automobile.
Again, thanks to Ian Sinclair for sharing his experience for your benefit.
What do you think? Let me hear your horror stories or your successes at Larry.English@infoimpact.com. I will pass your comments on to Ian.
Reference:
- See Larry P. English, "Check Digits as a Tool for IQ," DM Review, March 2003.
Larry P. English is president and principal of INFORMATION IMPACT International, Inc., Brentwood, Tennessee, and the author of the widely acclaimed book, Improving Data Warehouse and Business Information Quality: Methods for Reducing Costs and Increasing Profits. English is cofounder of the International Association for Information and Data Quality (www.iaidq.org). English is an internationally recognized speaker, teacher, consultant and author and may be reached at larry.english@infoimpact.com or through his Web site at www.infoimpact.com. For more on how to improve your IQ principles and techniques, and prevent your organization from wasting millions in information scrap and rework, join the IAIDQ (visit www.iaidq.org).
For more information on related topics, visit the following channels:


