This depends on the service and contingency.
E.g. gitlab lost issues and non-crucial things. Since it's basically git most of the data should still be there and recoverable. Users could always move to another host etc. So yes I'd use gitlab in theory although we currently use github.
Parse is an example of a service we shouldn't have used, since it relies on a fixed server that you can't move you have to push updates to all your mobile apps to migrate which would be a production nightmare. This applies to all backend-as-a-service solutions IMO.
Our mobile cloud solution is used in compile time, so in my (heavily biased opinion) it shouldn't be a problem. Even if the worst case scenario happens and we go down... We won't lose your data (we don't have it) & your customers will never know since the dependency on our service is during compile time, not runtime.