Trying to wrap my head about the following Viptela scenario

Hey guys,

I am not that savvy on Service Provider stuff, and I'm trying to study a little bit of Viptela (Cisco SD-WAN) using their documentation. The question is not really on the product itself but in the way private WAN's work, because the following section on " Allow Data Traffic Exchange across Private WANs" is throwing me off:

https://sdwan-docs.cisco.com/Product_Documentation/Software_Features/Release_18.1/04Segmentation/03Segmentation_(VPN)_Configuration_Examples_Configuration_Examples)

If I understand correctly, a data tunnel cannot be formed between two edge routers on the same private transport (MPLS, for example) IF THE NETWORK SEGMENT that connects the two routers to the transport is NOT advertised over the Service Provider network. This makes sense to me because if the two subnets are not advertised, both routers are not able to reach each other, fine.

The solution, however, as per the document, is to use a loopback interface instead? It reads:

" Because the loopback interfaces are advertised across the overlay network, the vEdge routers are able to learn reachability information, and they can exchange data traffic over the private network. "

But why does the loopback get advertised when the physical interface did not? The article doesn't explain well why by using a loopback, both routers are now able to reach each other's said loopbacks, and it does not mention those loopbacks being advertised through the MPLS network.

Am I missing something here?