Set Rules for Blank, Unknown, and Default Values During CRM Migration
Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.
A migration file contains an empty industry cell. The destination CRM already holds a verified industry value. Should the import preserve it, clear it, or replace it with a default? Unless someone makes that decision before import, the importer’s behavior becomes your data policy.
For CRM migration blank values and defaults, the safest starting rule is: do not let missing source information overwrite a valid destination value, and do not present an operational default as a confirmed fact. Exceptions need an explicit field-level rule, an authorised decision, and a test that proves the intended result.
Build a value-handling matrix before preparing the final import file. It should distinguish blank cells, explicit unknowns, unavailable fields, valid values, invalid values, and authorised clear instructions. For each state, specify the destination treatment, overwrite rule, provenance label, and expected validation result.
1. Separate source meaning from import mechanics
An empty-looking field can mean several different things. A blank cell may mean nobody entered a value. An explicit unknown means the source deliberately recorded that the answer was not known. An unavailable field means the export did not contain that field at all. None automatically means that a destination value is wrong.
Use a small state vocabulary before transforming the file:
- Present: a source value exists and passes the field’s validity checks.
- Blank: the field exists, but this record has no value.
- Explicit unknown: the source records a recognised unknown state.
- Unavailable: the field is absent from the source extract or was not collected by that system.
- Invalid: a value exists but cannot be accepted without correction.
- Clear requested: an authorised instruction explicitly requests removal of an existing value.
Keep these meanings separate from their technical representation. An omitted field, an empty string, and a null value may be handled differently by an import route. A token such as NULL could also be imported as literal text. Establish the behavior of the actual route you will use rather than assuming its interpretation.
Zero and false are not blank. An explicitly recorded count of zero is a value; an unanswered count is not. Likewise, a blank checkbox export does not establish false unless the source specification says it does.
2. Make overwrite decisions against destination state
A source-only mapping is incomplete. The decision also depends on whether the destination field is empty, populated, verified, or already carrying a known placeholder.
For each field, decide which source is authoritative and what evidence permits replacement. A valid incoming value might fill an empty destination but require review before replacing a verified one. Alternatively, a designated source system may own that field and be allowed to replace an older value. Neither policy should be inferred from which file happens to arrive last.
Use these baseline rules unless a documented exception applies:
- Blank, unknown, unavailable, and invalid source states do not overwrite valid destination values.
- A valid source value can fill an empty destination after format and allowed-value checks.
- Conflicting populated values follow the field’s authority rule or enter a review queue.
- A clear request uses a separate, restricted path rather than an ordinary blank cell.
- A required operational default is inserted only when the destination is empty and the record cannot proceed without it.
Do not use a timestamp alone to settle every conflict. A recent export timestamp says when the file was produced, not when the underlying value was confirmed. Record-level modification times can also change because an unrelated field was edited.
3. Use a filled value-handling matrix
Fictional worked example: a business is migrating account records into a CRM that already contains some records. In this hypothetical destination, industry is optional, lifecycle stage and owner are required, and a review status can distinguish records awaiting checks. These are example assumptions, not claims about a particular product.
The matrix below is a filled planning artifact. Each entry specifies the source state, destination treatment, overwrite rule, provenance label, and expected validation result. The labels describe a recommended migration log design; they are not assumed to be built-in CRM features.
VALUE-HANDLING MATRIX — HYPOTHETICAL A. Industry Source: Blank cell Destination: Existing verified value: Manufacturing Treatment: Do not write the field Overwrite: No Provenance: preserved_destination / source_blank Validation: Manufacturing remains unchanged B. Industry Source: Explicit token UNKNOWN Destination: Empty Treatment: Leave business value empty; log unknown state Overwrite: No write Provenance: source_explicit_unknown Validation: Empty industry; unknown state traceable C. Annual revenue Source: Column unavailable in export Destination: Existing populated value Treatment: Exclude field from write payload Overwrite: No Provenance: preserved_destination / source_unavailable Validation: Existing value remains unchanged D. Employee count Source: Numeric 0, valid under agreed source definition Destination: Empty Treatment: Write numeric 0 Overwrite: Empty destination only Provenance: source_present / source_record_reference Validation: Stored as 0, not blank or unknown E. Employee count Source: Text 'several' Destination: Existing numeric value: 12 Treatment: Preserve 12; queue source value for review Overwrite: No Provenance: preserved_destination / source_invalid Validation: 12 unchanged; invalid-value issue recorded F. Lifecycle stage — required Source: Blank cell Destination: Empty Treatment: Set Unclassified; mark review status Pending Overwrite: Empty destination only Provenance: migration_default / rule_stage_unclassified Validation: Required field populated; not treated as qualified G. Lifecycle stage — required Source: Blank cell Destination: Existing value: Customer Treatment: Do not write the field Overwrite: No Provenance: preserved_destination / source_blank Validation: Customer remains unchanged H. Owner — required Source: Field unavailable Destination: Empty Treatment: Assign authorised migration holding owner Overwrite: Empty destination only Provenance: migration_default / rule_holding_owner Validation: Owner exists; holding assignment is identifiable I. Website Source: Explicit authorised clear request Destination: Existing URL Treatment: Clear using tested import semantics Overwrite: Yes, for approved record and field only Provenance: authorised_clear / approval_reference Validation: Empty website; prior value retained in recovery log J. Industry Source: Valid value: Logistics Destination: Existing verified value: Manufacturing Treatment: Preserve destination; queue conflict Overwrite: Only after field-authority decision Provenance: conflict_pending / both_value_references Validation: Manufacturing unchanged; conflict remains visible
Replace the hypothetical field names and labels with your own. Add a separate entry whenever destination state changes the action. For example, “blank source plus empty destination” and “blank source plus populated destination” should not be collapsed into one ambiguous instruction.
4. Choose defaults that do not impersonate evidence
A required field creates a technical constraint, not permission to invent a business fact. If lifecycle stage is missing, defaulting to a sales-qualified stage can make the import look complete while changing how the record is interpreted.
Prefer a neutral, supported state such as Unclassified when the destination allows it. Confirm that reporting and workflow rules treat it as unresolved. A neutral label is not neutral in practice if a downstream process interprets every populated stage as qualification.
For each proposed default, document:
- Reason: which destination requirement makes it necessary?
- Meaning: is it a temporary operational assignment or a factual value?
- Effects: what reports, routing rules, or other processes consume it?
- Exit condition: what evidence allows replacement?
A holding owner can keep records assignable, but it also concentrates work in a queue. A shared Unknown category makes unresolved records easy to filter, but may appear as a real category in reports. Leaving a field empty preserves uncertainty, but may block record creation. Choose the least misleading supported option and document the cost.
If no safe required-field value exists, hold the affected records outside the production import until the schema or migration approach is resolved. A smaller, accurate import is preferable to fabricated certainty.
5. Preserve provenance without crowding business fields
Do not put explanatory sentences into numeric fields or invent an address merely to satisfy validation. Keep the business value separate from the explanation of how it arrived.
A migration decision log can hold the source record reference, destination record reference, field, source state, prior destination value, chosen action, rule identifier, batch identifier, and resulting value. Record source observation time where it is available and meaningful. Avoid creating a confirmation date from the migration date.
If the CRM supports suitable metadata fields, some provenance may live there. Otherwise, maintain a controlled companion log linked to the records. Verify available import and metadata options in the intended workspace and current plan; this workflow does not depend on a particular built-in control.
The important distinction is between source-supplied, preserved destination, migration default, and authorised clear. Someone reviewing a value later should be able to identify which category applies without reconstructing the entire migration.
6. Test the matrix, not just whether the file imports
A successful import message does not prove that the values mean what you intended. Test with a controlled batch containing every matrix branch, including populated destination records.
Hypothetical test plan: prepare one record for each matrix entry, plus separate cases for whitespace-only cells, a literal unknown token, an unmapped dropdown value, and a repeated import. Keep expected outcomes beside the test inputs.
- Inspect transformations. Confirm that trimming whitespace does not erase meaningful values and that unknown-token matching is field-specific.
- Inspect the prepared writes. Check that preserve actions omit the field rather than accidentally sending a clear instruction.
- Check stored values. Compare actual destination values with the matrix, including numeric zero and deliberately cleared fields.
- Check downstream effects. Review whether defaults alter assignment, reporting, or enabled workflows. Keep production actions paused where supported and authorised.
- Repeat the batch. Confirm that a second run does not overwrite later corrections or create new unintended changes.
Do not globally convert every occurrence of “N/A,” “none,” or “unknown” to blank. Their meaning depends on the field and source convention. An unrecognised token should become a review item, not an invitation to guess.
7. Define failure handling before the production run
Separate a rejected value from a rejected record. An invalid optional website may allow the rest of a record to migrate, provided the omission is logged. An unresolved required owner may block the entire record. State that policy in advance so different operators do not improvise different outcomes.
Keep a recoverable snapshot of affected destination values before writing, with access limited to the people who need it. Obtain authorisation for the import and for any bulk-clear operation. A clear instruction deserves stronger review than a fill-empty action because it deliberately removes information.
If a batch produces unexpected changes, stop subsequent batches and isolate the affected rule. Compare intended actions with actual stored values. Repair only the affected records and fields rather than replaying an entire file indiscriminately.
Recovery also needs conflict handling: a destination value may have been legitimately edited after import. Do not restore an old snapshot over that newer work without review. Retain the original rule version so the failure can be traced to a decision, transformation, or importer behavior.
8. Set a measurable release gate
Your next action is to complete the matrix for the fields that can cause the most damage: required fields, fields used for routing or reporting, and fields already populated in the destination. Expand it to the remaining mapped fields before production approval.
For a hypothetical pilot release gate, require every matrix branch to pass, no unintended clearing of populated destination values, and a recorded reason for every inserted default. Count unresolved conflicts, invalid values, blocked records, and defaulted fields separately rather than combining them into a single success percentage.
Finally, reconcile planned actions against observed outcomes: how many fields were preserved, filled, defaulted, cleared, or held for review? Investigate differences before expanding the batch. That comparison turns “the import completed” into a more useful answer: the destination now reflects the decisions you actually approved.
Put the next conversation to work.
Explore Alactic Communications for WhatsApp conversations, campaigns and AI-assisted workflows. Choose the product and plan that fit your team.
Explore Communications