Percolating before indexing VS Percolating after indexing


I understand that if we percolate before the document has been indexed then we just pass the document and run the required percolator and it gives the matching queries.

Now, what happens when the document has been indexed and I pass the doc id. Does Elastic first fetches the document from index and then passes it to the percolator ?

My question is whether percolating the document before indexing and adding some extra tags e.g query ids it matched more efficient than percolating it after indexing.

Also in case of multipercolate, I assume the Elastic iterates over each document one by one and then gives the consolidated response.
In this scenario also, will single percolation be more efficient than multi percolate.

Any pointers would be of great help.