Stephen - The Random Sample operations block can be inserted anywhere in the DataManager workflow...it currently requests sample size as row count rather than percentage of incoming row count but this would be simple to add, and is consistent with the way samples are typically described, e.g. 20% sample, 30% sample etc.

Note that we are talking here about random sampling, not Pareto-style top 20% filtering, i.e. filter out the bottom 80% by some criteria.