Memory hook: Match first, validate next, load in a recoverable batch.
Must remember
Choose Data Import Wizard or Data Loader according to supported objects, operation, volume and automation needs. Import Wizard provides guided supported imports; Data Loader supports broader bulk operations and mappings. Confirm current limits rather than assuming one memorized row count applies to every tool/edition.
Insert creates records; update changes matched records; upsert combines those behaviors using a supported match key such as an external ID. Validate key uniqueness, parent references and operation order. A wrong match key can create duplicates or update the wrong customer. Test a small batch and inspect success/error output before a full load.
Matching rules identify potential duplicates; duplicate rules decide the configured response, such as alerting or blocking. Validation rules reject data when their formula evaluates to true. Check how imports/API operations invoke automation and enforcement; bypass assumptions often cause bulk-load surprises.
Plan ownership transfers, mass deletion and archiving with related records and reporting needs in mind. Export/back up before consequential changes and verify that the recovery process includes relationships, metadata needs and a realistic restore procedure. A CSV export is not automatically a complete point-in-time org restoration.
Use consistent date/time, locale and currency handling. Protect exported personal/customer data and restrict access to error files that contain rejected record values. Reconcile counts, totals and sampled relationships after the operation. A successful job status can coexist with partially failed rows.
Choose under exam pressure
| Requirement | Choice and reason |
|---|---|
| Create new or update existing external records | Upsert with a validated external identifier. |
| Detect likely duplicate customers | Matching rule plus an appropriate duplicate-rule response. |
| Prevent an invalid business value | Validation rule with tested error behavior. |
Traps
- Validation rules block when the error condition is true.
- Bulk job completion does not mean every record succeeded.
Active recall
1. Update versus upsert?
Change matched existing records versus update or insert based on a supported key.
2. Matching versus duplicate rule?
Identify candidates versus decide what to do about a match.
3. Why import a sample first?
To validate mapping, relationships, automation and error handling.
4. Why preserve error files securely?
They may contain sensitive rejected record values.
5. What makes a backup useful?
A tested ability to restore the needed data and relationships, not merely possession of an export.