A business perspective may initially suggest that correctness is better since the information is used to drive important decisions however experience suggests consistency to be a far better focus as this drives trust and trust is something that is easy to lose but hard to gain.
Imagine if your car brakes worked inconsistently, would you have faith in them that they would work in an emergency? You may get them fixed but how long till you feel completely safe and trust them again? Conversely, if they were just becoming incorrect (a little dull and worn out) you would just get them replaced/fixed and all would be well again – you would always feel safe, not lose trust and you would be happy straight away.
We need to take account of the dichotomy of control and focus on the elements where we have true control; concerning ourselves with things internal to us (our databases and processes) and less so on external items where we have less control and normally less knowledge (the data itself).
We can control the end to end data process but when it comes to the actual data we have a lot less control since there is normally some kind of human intervention in how it is created.
So to answer the title of this blog we should put most of our effort into the process of ensuring that the information is consistent. Our coding should ensure the processing is not just right but more that it is never wrong. We shouldn’t ignore the data but rather put controls in place to flag up potential issues for further investigation – it’s quite likely that a business process has changed and we are unaware of this change.