First step is to check if the drop in usage activity is a legit drop in itself.
Break the data down into chunks and see if there’s a missing datapoint, confirm with data engg teams that the pipeline is working well and once we are sure everything is working well, will start triangulation on what all characteristics we can pinpoint for the drop.
Once we have exhausted quantitative options then go for qualitative options. These questions are never right or wrong.
I work as a B2B SaaS PM. Let me tell you, CSMs and Support team are usually clueless about the larger picture. They only work in silos and only on their accounts not having an overview of whats going on overall so their inputs would also be quite qualitative. 🤪