As we know so well, at the end of a retro you decide at least one item to continue to improve your process. I expect that on a future retro you will evaluate whether you managed to implement that item and whether doing that made any difference.
Why not borrow a page from vaccine development manual or any scientific investigation?
When you conduct experiments, you need to have an hypothesis and then you need to decide how to measure whether the hypothesis was true.
For example, your hypothesis is that “swarming” will increase flow of the backlog items thru the system. Yes, I know I picked one that is easy to prove…
You could measure this by having some team members swarm while others do not or you might be able to compare between teams (as long it is not comparing things like velocity!!!! DON’T DO THAT, EVER!!!)
Or you could compare the number of bugs created by a team that does not do code review versus one that does.
Only with a hypothesis and an plan to measure results you will validate whether your hypothesis was true.
Speaking of vaccine development, one characteristic of scientific studies is that they are published and yes, I know that what happens during the retro stays with the teams.
However, if you discover something cool that other teams and the organization can benefit from it, why not tell the other teams?
When other teams learn about your experiments, they will want to reproduce them and it will lead to great things or they may find that you did some local optimization.
Either way, we all will benefit from what we share.