One thing that often gets forgotten in the discussions about whether to soft delete and how to do it is: what about analysis of your data? Even if you don't have a data science team, or even a dedicated business analyst, there's a good chance that somebody at some point will want to analyze something in the data. And there's a good chance that the analysis will either be explicitly "intertemporal" in that it looks at and compares data from various points in time, or implicitly in that the data spans a long time range and you need to know the states of various entities "as of" a particular time in history. If you didn't keep snapshots and you don't have soft edits/deletes you're kinda SoL. Don't forget the data people down the line... which might include you, trying to make a product decision or diagnose a slippery production bug.