Audience question: which approach is best when staff members and admins sometimes need aggregated computations across multiple tenants? Answer: both schema-level and row-level approaches allow this — you can join across schemas by prefixing table names (tenant_a.table_name) or simply drop the tenant_id filter. Row-level is easier here because it just means extending the WHERE-clause. Schema-level makes per-tenant backup easier but aggregation across everything harder. Row-level is therefore the recommended default — also used at Shopify and Salesforce.