For a Data Mapper Extract running against a very large Salesforce object, the critical design concern is query selectivity. A filter on PolicyNumber__c must be supported by an indexed field so Salesforce can avoid a full table scan across millions of rows. Marking the field as an External ID creates an index-supported lookup pattern, and requesting a custom index is also a valid large-data-volume design approach. Making the field required does not make the query selective. A Text Area field is worse because long text-style fields are not appropriate for selective indexed filtering. Data Mappers do not magically solve poor physical data-model design; they still depend on Salesforce query performance rules. Salesforce large-data-volume guidance recommends indexed/selective filters for scalable queries.

================