Thats indeed something that doesn't seem to be discussed much, even if you look at graphql's presentation - https://www.youtube.com/watch?v=WQLzZf34FJ8 https://

Thats indeed something that doesn't seem to be discussed much, even if you look at graphql's presentation - https://www.youtube.com/watch?v=WQLzZf34FJ8 https://www.youtube.com/watch?v=WQLzZf34FJ8 there seems to be a lot of redundancy when "my friends including their events" is fetched. I imagine that a scheme based on an UUID and deduplication would work - when constructing the response an object repository is kept and if an object with the same uuid appears twice its replaced with a reference to them. Going against the "don't extend JSON" best practice, that would be: { "friends": [{ "name": "Name", "events": [#uuid1, #uuid2] }, { "name": "Another name", "events": [#uuid1, #uuid2] }], } #{ "uuid1": {"name": "MyEvent", "description": "..."}, "uuid2": {"name": "OtherEvent", "description": "..."}, } And it can support cyclic references too, as a bonus. Ignoring best practices is so... liberating. Ultimately this is just compression though. Its possible that gzip will be quite sufficient for most cases.