I am constantly reminded in working with clients doing data capture that the cost of error is not well understood or dealt with effectively. For example, in one case, a single letter error on a stock certificate (which only costs about a penny to capture correctly) actually cost an organization $274.50 to correct the error. That’s a cost-to-fix/cost-to capture ratio of 27,450 to 1!
One problem apparently is that the cost of error may accrue in a different organization than the one doing the data capture, and so it may be easily overlooked. Another problem is that “Nobody’s Complaining”, which is right up there with “Not My Job”. Another classic problem is that if good testing is not employed in set-up and production, the errors will not be found, at least by the data capture group.
Good testing approaches are required, and improvement processes put in place to realize these savings, but they are really worth it. There are efficient testing approaches that are largely automated for speed, efficiency, and low cost. In addition, many improvement techniques are available to make your data capture better. Improvement ideas include better form design, use of color dropout, understanding field vs. character recognition, use of context, dictionaries and tri-grams, etc.
In my book, Handprint Data Capture in Forms Processing, for example, I give more detail on data capture based on 18 years experience with the U.S. Census Bureau and other agencies, including cost of error, improvement methods, statistical sampling, efficient testing techniques and more. If you would like to peek at some more free hints about all this, you can “look inside the book” on Amazon.com at:
Brad Paxton#ScanningandCapture #CostofError #DataQuality #testing #Capture