Audience question asking whether the team evaluated data denormalization (e.g. adding a tags field directly on the Post model) instead of, or in addition to, graphql-batch and graphql-preload. Speaker answered that graphql-batch was the first tool they introduced, the other examples were simplified, and because Check's data model is large and no single field dominates, they handle performance mostly at the query level and per-field rather than via denormalization.