This seems like a super important observation. If the number of dataset saw a massive jump (50%), then back down a week later, that seems more like the correct

This seems like a super important observation. If the number of dataset saw a massive jump (50%), then back down a week later, that seems more like the correction of an error.