Return to current issue.
Bad Data Can Cost You Big Time
By Ira Kaplan
Bad data is pervasive; it infiltrates paperwork and databases. Seemingly innocuous, it derails multi-million dollar computer systems, leads to lost revenue, and erodes customer goodwill. This article examines organizational disasters with data as the leading culprit, defines data quality, illustrates the cost of bad data, and shows where improved data quality yields measurable benefits.
We work with data in the form of catalogues, purchase orders, invoices, shipping statements, proof of deliveries, remittances, return authorizations, credit and debit memos, and checks. Dr. Tom Redman, statistician, author, and president of Navesink Consulting Group, defines data as "the facts and figures associated with customers, products and services, market and financial performance - indeed, every aspect of life in the information age." Some data is of poor quality and hides in our daily work, e.g., unconverted units of measure, extra zeroes in a spreadsheet cell, and stale data. Much of this is accepted as unavoidable, the cost of doing business. Yet a small error can result in disaster. Let's look at a few stories.
Lost in Space: In 1999 NASA's Mars Climate Orbiter, a $125 million spacecraft, was lost when attempting to orbit Mars and was never found. Lockheed Martin, a NASA subcontractor, did not use the metric measurement system as specified by NASA, inches and pounds were not translated into meters and kilograms. Dr. Edward Stone, director of the Jet Propulsion Laboratory, summed it up best: "Our inability to recognize and correct this simple error has had major implications." The entire project cost $327 million dollars.
CD Mail Fraud: David Russo, 33, of Sayreville, NJ swindled two music mail-order clubs, BMG Music Services and Columbia House Music Club, of 22,260 compact discs for four years. He used 1,630 aliases to purchase the music CDs at prices reserved for first-time buyers and evaded the music clubs' computerized anti-fraud measures by making each mailing address slightly different. Tactics included adding sham apartment numbers, unnecessary abbreviations, and extra punctuation. Selling the compact discs at flea markets for about $10 each yielded nearly $167,000 in profits.
Extra Zeroes: In 2005 Eastman Kodak Co. increased its third-quarter loss by $9 million due to several accounting errors. Robert Brust, Kodak's chief financial officer, said the severance-related error stemmed from miscalculating severance pay accrued by just one employee, exposing "an internal control deficiency that constitutes a material weakness that impacted the accounting for restructurings." A spokesman further explained it as a faulty spreadsheet calculation; too many zeroes were added to the employee's accrued severance. Kodak caught the error in time so no payment was made.
Defining Data Quality
Dr. Joseph M. Juran, co-author of the highly regarded "Juran's Quality Handbook" states "Data to be of high quality if they are fit for their intended uses in operations, decision making and planning."
What makes data fit for use? Of hundreds of data quality characteristics which can be evaluated, accuracy, timeliness, completeness, and consistency have proven to be useful. Set measurable quality goals for each attribute from the perspectives of the intended use and the intended user or data consumer. For example, auditing requires accuracy to the penny while a corporation's financial position can be stated to the nearest thousand dollars.
The Cost of Bad Data
Data errors do not have to trigger disaster to impair a business. Redman conservatively estimates poor data quality costs a typical company ten percent of revenue and indicates 20 percent is more likely. He and his associates have observed that most organizations find and correct data errors in the normal flow of work and on occasion perform a massive data cleaning. They developed a "Rule of Ten" to estimate data cleaning costs: if it costs $1.00 to complete a simple operation when all the data is perfect, then it costs $10.00 when it is not, i.e., late, hard to interpret, incorrect, etc.
Companies often overestimate their data quality and underestimate revenue lost to data errors and inconsistencies. Wayne Eckerson, director of education and research for The Data Warehousing Institute (www.tdwi.org), surveyed businesses and almost 50 percent of respondents stated their data quality is good or excellent. But more than 33 percent of respondents stated their data quality is "worse than the organization thinks."
Is Fixing Bad Data Worth the Effort?
Fixing bad data requires significant and ongoing effort. Is it worth the cost in manpower and capital resources?
Manufacturers and vendors are benefitting by verifying their products' weights and dimensions against published standards. Wegmans supermarket chain underestimated truckload volume due to years of uncertain data accuracy in their software system. With newly verified data they are able to use 18 percent more truck capacity and save 100 shipments per week from distribution center to store.
More examples:
- One year after automating product data synchronization with retailers, Clorox decreased invoice deductions by 40 percent.
- A manufacturer corrected the recorded weight of a single item and saved $2.2 million in annual transportation costs.
- By reducing through automation the number of manual touches on product information a manufacturer reduced annual fines, deductions, imposed by retailers from $300,000 - $500,000 to almost zero.
- Sharing accurate item data with trading partners, a manufacturer improved order and invoice administration and reduced inspection time five minutes per order. The time savings improved productivity by estimated 59,000 hours annually or 29.5 man-years!
Where to get started?
Chances are good your business has an overwhelming volume of data, all worth inspecting for shortcomings. But not all data has the same value to your business and cleaning data requires significant effort. So where does one begin? Where is the potential payback greatest? A good start is with data that touches partners and customers. Below are some examples.
The Advanced Ship Notice (ASN), delivered through Electronic Data Interchange (EDI), is a worthwhile starting point. According to a study by the Vendor Compliance Federation and GXS, improving data accuracy between trading partners "will allow for unimpeded product flow that contributes to a high velocity supply chain for better use of financial assets." The ASN is sent from the supplier to the retailer, notifying the retailer that items from one or more purchase orders have been shipped. An accurate ASN enables in-shipment tracking and automated receiving, saving both time and cost. The ASN contains data from purchase orders and invoices; the data quality of these source documents impact the ASN. This is often the case when embarking on a data improvement project for one document, database table, or process: it leads you to the next one to be improved. For more information, see "Business 4 Business" at www.vcfww.com.
In 2002 The Data Warehousing Institute (TDWI) estimated "that poor quality customer data costs U.S. businesses a staggering $611 billion a year in postage, printing and staff overhead." They were referring only to name and address data and not the myriad other data elements captured in a typical customer relationship management system. Wayne Eckerson, TDWI's director of education and research, says the problem is much larger than postage, printing, and staff costs. Alienating and eventually losing loyal customers through incorrectly addressed letters or failing to recognize a frequent customer when she calls, visits the store, or surfs the Web site is a loss of sales and referrals. Eckerson suggests processes for improving customer data at http://www.adtmag.com/article.aspx?id=6321.
Credit and collections departments, often under-supported by their companies' information technology groups, often end up using manual followups and Microsoft Excel to fill the gaps left by their financial accounting systems, in order to track and reconcile invoices and deductions. The best solution is an application program built on top of a relational database with a Web browser as the graphical user interface. This model, Software as a Service (SaaS), is successfully used by Salesforce.com (www.salesforce.com) for sales force automation, 37signals (www.37signals.com) for easy-to-use lightweight applications for project management and customer relationship management, and Smyth Solutions' Cfari (http://www.smythsolutions.com/smyth_cfari.asp) for accounts receivables management focused on workflow, deductions management, and reconciliation.
Ira Kaplan is Director of Cfari Product Management at Smyth Solutions. Cfari™ is an accounts receivable suite optimized for corporate credit, collections, deductions, reconciliation, and cash application. Cfari is implemented using the software as a service (SaaS) Web-based model for companies needing a flexible accounts receivables solution either as the system of record or as a powerful bolt-on for specific needs, such as invoice collection campaigns, deduction resolution and mass account reconciliations.
References
- The New York Times, September 25, 1999. National News Briefs; Hope for Orbiter Ends
- Mars Climate Orbiter Fact Sheet - http://mars.jpl.nasa.gov/msp98/orbiter/fact.html
- NASA News Release 99-113, Mars Climate Orbiter Team Finds Likely Cause of Loss, http://mars.jpl.nasa.gov/msp98/news/mco990930.html
- NASA Press Release: 99-134, MARS CLIMATE ORBITER FAILURE BOARD RELEASES REPORT,NUMEROUS NASA ACTIONS UNDERWAY IN RESPONSE, http://mars.jpl.nasa.gov/msp98/news/mco991110.html
- Enterprise Knowledge Management: The Data Quality Approach, by David Loshin (Author), 491 pages, Publisher: Morgan Kaufmann; 1st edition (January 22, 2001)
- NY Times March 25, 2000 "Man Admits Fraud in Joining CD Clubs Thousands of Times"
- Talking Quality With – Joseph M. Juran; The Guru of Doing It Right Still Sees Much Work to Do. By Claudia H. Deutsch New York Times, Published: November 15, 1998
- Data: An Unfolding Quality Disaster, Thomas C. Redman, DM Review Magazine, August 2004
- Dow Jones' MarketWatch, November 9, 2005: Kodak restates, adds $9 million to loss. Dow Jones Market Watch (www.marketwatch.com)
- Data Quality Assessment Leo L. Pipino, Yang W. Lee, and Richard Y. Wang, Communications of the ACM April 2002/Vol. 45, No. 4ve web.mit.edu/tdqm/www/tdqmpub/PipinoLeeWangCACMApr02.pdf
- Application Development Trends. Data Warehousing Special Report: Data quality and the bottom line. 5/1/2002 By Wayne W. Eckerson http://www.adtmag.com/print.aspx?id=6321
- Business 4 Business: ASN and Data Accuracy Drive Higher Order Functionality and Redefinition of Retail B2B Communications. November 2007, A VCF & GXS White Paper. http://www.gxs.com/forms/0711_B4B_wp_VCF.htm
- Synchronization - The Next Generation of Business Partnering. http://www.transora.com/data_accuracy.html
Return to current issue.
