The vast majority of the data your plant has produced is likely gone.

Data is deliberately deleted either through aggregation (keeping the average of many values) or hard deletion. The cost of storage decreased by a lot over the last decade (I still remember paying 50 EUR for a 50 MB PlayStation 2 memory card).

Sometimes, data is deleted by accident: a server runs out of disk space without anyone noticing. I’ve seen projects where important data was identified only to discover that nearly all of the data older than six months was lost.

A third, more insidious, way in which data is lost is a missing context. The data is still on disk, but the context is not clear. Some signals changed names, others were added or removed. Maybe at some point the machine was revamped. The older the data, the more likely its context got lost.

When a company boasts they track lots of data its always good to ask yourself how much of that data they still have and how much of it is clear.