Dependent data sources wrongly requiring a re-read

Our Terraform codebase is relatively large, with the topmost resource of our graph being an aws_organizations_organization resource. When there are any changed to this resource’s outputs, eg creating new OUs or it has discovered a new account in the org, as it is the top resource, it triggers rereads of all of the data sources in the codebase. I naïvely tried adding lifecycle block to ignore changes to the outputs of the resource, but that didn’t stop ~500 changes being planned.

Is it possible to avoid this by changing either our configuration or the provider code, or is this a core Terraform problem?