There were some teams that never stopped coming to the office, due to sensitive aspects of their work.

And there's a boatload of data from centralized project management and ticketing systems, as well as centralized source repositories.

The data was absolutely available for data-driven arguments.

Unfortunately you'd probably have to discount the teams that never switched to WFH. The same reasons that likely drove that decision would mean they're unlikely to be good comparators to teams that had a choice.

So what you are saying is “there’s no possible way to collect data on this, so just trust me even though I’m bias”

That's a lot of words to shove in my mouth. At least buy me dinner first.