Here's where it gets really simple. Resize the staging instance to match live. Put live into maintenance mode and begin the data transfer to staging (with a lot of cloud providers, step #1 and #2 can be done in parallel). As soon as it finishes copying, take live down, point the DNS records at staging and wait for a few minutes. Staging is now live, with all of live's data. Problem solved. Total downtime: hardly anything compared to not being prepared. Total dataloss: none.