The Worth of Data

If the data is not statistically significant, it is worse than worthless.  It has negative worth.  If you do not act on the data, it is worse than worthless.  It has negative worth.

By negative worth I mean that you have cost the system money, but received nothing in return.  So, how can we get more from our data?  Here we note that analog is better than counting that is much better than the next to worthless binary data.  Unfortunately, binary data is what we normally collect when we say a wound roll is either good or bad or it has a customer complaint or it does not.

Let’s be more specific with an example.  A single wrinkle in a wound roll may render it unusable with certain products and customers.  Thus, a roll with 10 or 100 wrinkles has the same saleable value as one that has only one wrinkle.  Yet, you are clearly far closer to success with just one wrinkle per roll than 10 or 100.  So counting wrinkles in a roll versus counting rolls with wrinkles allows your sample size to go way down or your confidence to go way up for the same roll or run.  More bang for your buck, so to speak.  We can get even more worth from the same roll or run if we were to take measure of the accumulated length of the wrinkles.  In that case, we have analog data as a response.  A little more analysis work for sure, but probably a lot less costly than increasing the number of rolls you need to run to have data that has value.  Finally, you can show progress to the boss who is looking for a home run when both of you should be satisfied (for the moment) with a solid base hit.

If you like my work, please consider joining my AllWebHandling group on Linkedin.