Decide Which Customer Fields Are Worth Keeping
Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.
A customer record has a field called “Customer category,” another called “Segment,” and a third called “Account type.” Each was added for a reason. Nobody can explain whether those reasons still apply, but everyone worries that removing a field will break something.
To decide which customer data fields to keep, connect each field to a named operational decision, verify how it is actually used, and identify the least detail that decision requires. Then choose one action: keep, simplify, stop collecting, or remove. Test the change against real work before making an irreversible deletion.
The useful question is not “Could this information be valuable?” Almost anything could be. Ask instead: “Who uses this value, for which current task, and what would they do differently without it?” The worksheet below turns that question into a review you can run on an existing customer record structure.
1. Give every field a specific job
A field earns its place by supporting an identifiable action, calculation, routing choice, or necessary record lookup. “Reporting,” “personalisation,” and “future use” are not specific enough to evaluate.
Write a purpose statement in this form: [Role] uses [field] at [trigger] to decide [action]. For example, a fictional support team might use preferred service language when assigning a new request to an available agent. The test is whether changing the value changes the assignment.
Separate the purpose from the place the value appears. Being displayed on a dashboard does not establish that anyone acts on it. Equally, a field that employees never open may still be necessary if an active rule uses it.
- Strong purpose: The account coordinator uses the assigned owner when a request arrives to identify who must respond.
- Weak purpose: Ownership information helps us understand customers.
- Unproven purpose: Someone may want an ownership report later.
Do not force an answer when nobody knows. Mark the purpose as unverified, name someone to investigate, and set a review date. Uncertainty should become a task, not a permanent exemption from review.
2. Check actual use before accepting a claimed dependency
Ask the field’s proposed owner to show the task, report, rule, or handoff that uses it. A recent example is more useful than a statement that the field is important.
Review both human and system use. Relevant places include saved views, reports, exports, assignment rules, message-selection logic, calculated values, and recurring spreadsheet work. Include connected processes outside the record screen; a field can look unused in the CRM while feeding a recurring operational file.
Where usage records are available, use them as evidence, not as the sole verdict. A frequently displayed field might be ignored. A rarely used field might support an infrequent but important task. Choose an observation period that includes the relevant business cycle rather than treating a quiet week as proof of redundancy.
For each dependency, record what it needs:
- Presence: Does it only check that a value exists?
- Meaning: Does it rely on a specific category or interpretation?
- Format: Does it expect a date, number, or exact text value?
- History: Does it require earlier values rather than just the current one?
Check identifiers and status fields especially carefully. A label that looks administrative may connect records or control a workflow. Do not infer that it is disposable merely because staff do not edit it.
3. Choose the action and the minimum useful precision
The review should produce a decision, not simply a cleaner field description. Use these four actions consistently:
- Keep: A current, demonstrated purpose requires the field at its present level of detail.
- Simplify: The purpose is valid, but fewer categories, less precision, or one authoritative field would support it.
- Stop collecting: New values are unnecessary, but existing values still serve a bounded historical purpose. Specify when that remaining use will be reviewed.
- Remove: No necessary current or historical use remains, and dependency checks and testing support removal.
Hiding a field is not the same as stopping collection. Imports, background processes, or other entry points may continue to populate it. Likewise, stopping collection does not remove existing values.
Precision deserves a separate decision. If a team only needs a service region, a full address may add unnecessary detail. But if another active process needs the address for a specific task, replacing it with a region would disrupt that work. Review every named purpose before reducing detail.
Do not merge fields just because their labels sound similar. “Account owner” and “Current request handler” can represent different responsibilities. Simplification is useful only when it preserves the distinctions people actually need.
4. Use this filled fictional field-review worksheet
Fictional worked example: The following worksheet describes an invented service business reviewing its customer records. All roles, workflows, field values, dates, and test thresholds are hypothetical. The dates illustrate a review schedule, not a prescribed deadline.
Field: Assigned account owner
- Purpose: The service coordinator identifies who is accountable when a customer request arrives.
- Actual use: Coordinators consult the owner in daily handoffs; an unassigned-account view is checked during workload review.
- Required precision: One current accountable person, with an explicit unassigned state. A department name is insufficient.
- Downstream dependencies: Handoff view, ownership report, and assignment exception queue.
- Owner: Service operations lead.
- Decision: Keep; clarify that this is account accountability, not the handler of every request.
- Removal test: Walk through an incoming request without the field in a test view. If no other authoritative source identifies accountability, removal fails. Confirm that reassigned accounts appear under the correct owner.
- Review date: 15 September 2026.
Field: Customer category
- Purpose: The service coordinator chooses between standard and specialist handling.
- Actual use: The handling guide uses only those two paths, although the field contains several legacy categories.
- Required precision: Standard, specialist, or needs review. Preserve unresolved cases rather than guessing.
- Downstream dependencies: Handling guide and a saved specialist-work view.
- Owner: Service delivery manager.
- Decision: Simplify after documenting how each old value maps to the new choices.
- Removal test: Apply the proposed mapping to a test copy. Compare old and proposed handling decisions, including blank and unfamiliar values. Investigate every changed route before release.
- Review date: 22 September 2026.
Field: Original campaign code
- Purpose: The marketing analyst compares acquisition cohorts in an existing historical report.
- Actual use: Existing values support that report; current acquisition work no longer uses this coding scheme.
- Required precision: The original code for historical records. No new code is needed.
- Downstream dependencies: Historical cohort report and its export.
- Owner: Marketing operations lead.
- Decision: Stop collecting; retain existing values while the named report remains in use.
- Removal test: Test a new record with no code through current workflows. Separately verify that the historical report still reads existing values and does not assume every new record has a code.
- Review date: 30 November 2026, after the next cohort review.
Field: Customer birthday
- Purpose: Previously proposed for birthday messages; no current operational purpose identified.
- Actual use: The proposed activity did not become an active workflow. The review finds no current report, rule, or manual task using the field.
- Required precision: None.
- Downstream dependencies: An obsolete export still includes the column, although its recipient does not use it.
- Owner: Customer operations lead, responsible for retiring the field.
- Decision: Remove after changing the export and completing the test.
- Removal test: Exclude the column from a test export and stop requesting new values. Have the receiving team complete its normal task. Check that no active process restores the value.
- Review date: 29 September 2026.
Field: Preferred service language
- Purpose: Support staff choose a suitable language for service responses.
- Actual use: Staff consult it before responding, but free-text variants make filtering unreliable.
- Required precision: A maintained language choice plus unknown; do not infer language from a customer’s name.
- Downstream dependencies: Agent views and the manual response-selection process.
- Owner: Support team lead.
- Decision: Simplify the entry format without discarding unresolved values until reviewed.
- Removal test: Test the proposed choices against varied existing entries. Confirm that ambiguous entries become review cases rather than silently receiving an assumed language.
- Review date: 22 September 2026.
Field: Contact restriction status
- Purpose: Staff determine whether a proposed outbound contact must be withheld.
- Actual use: Contact preparation includes a restriction check.
- Required precision: Enough to distinguish allowed, restricted, and unresolved situations for the relevant channel; a generic active-customer flag is not a substitute.
- Downstream dependencies: Contact preparation checklist and exclusion review.
- Owner: Communications operations lead.
- Decision: Keep pending a separate review of how restrictions are recorded and applied.
- Removal test: Use fictional restricted and unresolved records in a preparation-only test. Confirm they are withheld from the proposed contact list. Do not send messages as part of the test.
- Review date: 15 September 2026.
5. Test absence before deleting values
The safest first experiment is usually reversible: stop a selected collection path, exclude a field from a test view, or run a process against a test copy without it. These are recommended workflow designs, not claims that a particular product provides those controls. Check the capabilities and permissions available in your workspace before choosing a method.
Write the pass condition before running the test. “Nobody complained” is weak evidence. Better observations include whether requests reached the correct owner, reports reconciled, exports were accepted by the receiving process, or staff completed the task without reconstructing the missing value elsewhere.
Use representative cases: a normal record, a blank value, an unusual value, and a record affected by each known dependency. For a simplified category, compare the decisions produced by the old and proposed structures. For a removal, demonstrate that the necessary task still completes without the field.
Keep three states separate: the proposed decision, the completed test, and the implemented change. Approval to investigate removal is not approval to delete historical values.
6. Watch for false success and recover deliberately
A field review can appear successful while merely moving work out of sight. Ask staff whether they created private notes, spreadsheets, or repeated customer questions to compensate. Those are signs that the original purpose was missed or the replacement is inadequate.
- A rare task breaks: Extend testing to the relevant operating cycle. Restore the field’s use if needed while documenting the missed dependency.
- Categories were merged incorrectly: Pause the rollout and correct the mapping. Do not overwrite unresolved distinctions merely to finish the cleanup.
- New values keep appearing: Check other collection paths and connected processes before declaring collection stopped.
- An old report fails: Determine whether the report still supports a current decision. Repair a necessary dependency; retire an obsolete report rather than preserving its inputs indefinitely.
Before implementation, agree who can pause the change and how recovery will work. A temporary rollback copy may be appropriate for an authorised test, but it needs a named custodian, restricted access, and a defined disposal decision. Creating an indefinite shadow archive undermines the review.
7. Close each review with an observable result
Count completed decisions separately from implemented changes. A field marked “remove” is not removed if it remains in imports, exports, and staff instructions. A simplified field is not working if people still depend on the old values.
For the fictional worksheet above, the next step would be to assign the named owners, confirm the dependencies, and schedule the tests before changing the live structure. After each test, record the result, unexpected effects, and final authorisation.
For your own first review, choose a small group of disputed fields from one operational workflow. Fill every worksheet heading, including a named owner and review date. Measure how many have a demonstrated purpose, how many have an unresolved dependency, and how many changes pass their predefined test. That produces a defensible reduction plan—not just a shorter record screen.
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