Best practice is to create MZ or TAG rules for the Process Group. You can then specify that propagation to services and hosts should occur within this rule.
This is best practice because when you configure a rule per service for example, it may not always work because the service may not always be active.
Of course, additional rules may need to be added - however, in such a case as you have written this rule on Process Group should suffice.
You can try to tag processes where host group is XXXX and mark on those flags. Then, all hosts, processes and services with that host group would be tagged.
And then, you can create a management zone where processes, hosts and services have that tag.
In general, Dynatrace recommends using Host-based tagging rules as little as possible. For large environments, this results in a large number of configuration variables that you have to maintain. As I mentioned focus on defining tags based on Process Groups and extend tag rule to services and hosts. Process Groups are the key building blocks of most modern web applications.
It is not possible to propagate tags from a host to a service because in our model the hosts are on the infrastructure layer.