How to reduce execution time of GraphQL data fetching

If I start with a large GraphQL query that has 600+ spans on the dataFetching graphql resource, and I extract part of the query so that there are only 200 spans on that resource, the execution time of dataFetching doesn’t change at all. I don’t really understand why that is? If it’s not fetching that data, should the execution time be shorter? Maybe I have a lack of understanding how the data fetching work? I’m seeing this in DataDog when comparing execution times for a query before and after the change when looking at a stack list of specific traces, and any insight would be helpful!