Data intake, governance and exit handover
Plan intake, updates, retention and exit with clear ownership of sources and knowledge.
Inventory data before import
Record owner, format, volume, sensitivity, purpose and location for each dataset. Identify the source of truth and distinguish indexes, references, working copies and approved knowledge in DigitalHe. Sample scans, historical sheets and cross-system field differences before estimating cleaning effort.
Agree transformation rules
Agree field meanings, timezones, currencies, null handling and deduplication. Put ambiguous data in an exception list for owner review, preserving original values and transformation records.
- Identifiers: how to recognize the same project, customer or business event.
- Updates: full or incremental, deletion propagation and conflict ownership.
- Reconciliation: counts, material totals and sampled records agree.
- Recovery: checkpoints, duplicate-import checks and data rollback.
Approve model data exposure
Approve which fields each task may send to external models, redacting or aggregating where needed. Local storage does not rule out external inference. The customer owns approvals for provider retention, transfer routes and internal policy; the delivery team documents actual data flows.
Cover all copies in retention and deletion
Inventory source files, databases, indexes, logs, caches, backups and third-party copies, each with a retention period and owner. Confirm backup expiry and provider policies after deletion. Do not claim complete erasure without verifying the relevant copies.
What to hand over at exit
Agree exportable datasets, formats, references and handover at kickoff. Scope migration scripts or manual preparation where the current version lacks suitable export. At exit, hand over configuration records, customer data, agreed custom work and open issues, then revoke temporary identities. Contracts distinguish third-party licenses and general platform intellectual property.