Comparison
Data qualityvsData quality dimensions
Data quality
nothing errored, every job is green, and the number on the board is wrong by eleven percent.
Whether data is fit for the decision someone is about to make with it. It is deliberately a relative standard: the customer table that is perfectly adequate for a monthly trend is unusable for sending letters. Treating it as an absolute produces either endless tests nobody reads or none at all, so the useful version always starts with what the data is for and who is harmed when it is wrong.
Full entry →Data quality dimensions
you split 'the data is bad' into six specific claims, and it turns out only one of them is actually true.
The standard axes for saying what is wrong: completeness, accuracy, consistency, timeliness, validity, uniqueness. Their value is entirely in decomposition — 'bad data' is not actionable and 'twelve percent of rows have a null postcode' is. Each dimension also maps to a different test and a different owner, which is what makes the vocabulary worth having in a meeting.
Full entry →