(1) Pencil-whipping. "Checking the box" that the product passed rather than specifically showing what the measurement was.
(2) Lack of training for tools. One plant manager learned that over half of his folks did not know how to use a measuring tape even though he expected them to complete measurements on their products every 2 hours.
(3) Having operators simply regurgitate what the label says measurements should be rather than measuring the product to confirm that the data was actually "as specified."
(4) Accessing the data. I may have a database that tells me that I had 11 rejects off of a line, but it doesn't tell me what those rejects are. I may have another database that tells me what all the rejects that I had were... but it doesn't tell me the line that they came off of.
This last one is a HUGE struggle when it comes to big data. You lost 10% of your product, but you don't know why... or, you lost 10% due to scratches, but you don't know what product they were on. This is a REAL problem when trying to do root cause analysis.