Best practice for GraphQL data fetching in root vs page-level queries

The community member is asking whether it's a good practice to include full GraphQL queries (with rich content like long ACF text fields) in root-level data fetches, or if it would be more optimal to only fetch essential metadata (like IDs, slugs, and short titles) in the root, and then fetch full content on page-level only when needed. The concern is about performance when the same root query runs across many pages.

In the comments, another community member understands that if a query is added to the root, it will be done on all templates. The response confirms this, and mentions that in the future, usage detection will be improved.