Back to all articlesData stewardship

Stop Collecting Customer Data That Nobody Uses

Alactic Editorial10 min read

Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.

The customer record asks for a department name. Sales says it never uses it. Operations thinks a report might depend on it. Nobody wants to approve removing the field because nobody can explain what would break.

To stop collecting unused customer data safely, treat the change as a bounded operational test: identify likely dependencies, pause new collection for a defined group, watch the processes that could fail, and require evidence for exceptions. Then record a decision to retire collection, restore it for a specific purpose, or extend the test to cover an unresolved dependency.

This is not a project to delete historical values or redesign a form. Existing records can remain unchanged while you establish whether future collection earns its place. The goal is not to prove that nobody has ever looked at a field. It is to determine whether a current, necessary process depends on collecting it.

1. Turn “we might need it” into a testable claim

A field can be familiar without being useful. It can also have a genuine dependency that its nominal owner does not know about. Neither a quick staff poll nor an empty usage report is enough to distinguish those situations.

Start with a short purpose statement: who uses the field, for what decision or task, and when? A useful answer names an observable action. “The onboarding coordinator uses department name to assign the implementation contact” can be tested. “It helps us understand customers” needs more detail before it justifies continuing collection.

Ask anyone requesting continued collection to complete this sentence:

Without this value, this role cannot complete this task at this point in the process; the observable consequence is this, and the available alternative is this.

Do not treat uncertain answers as proof of waste. Put them on the dependency list and assign someone to investigate. Equally, do not let uncertainty become an indefinite veto. Each unresolved claim needs an owner, a check, and a decision date.

Exclude fields whose purposes or obligations have not been established well enough to test safely. For example, do not casually pause permission or opt-out information because a commercial team does not consult it. Resolve the relevant requirement with its responsible owner before including it in a collection experiment.

2. Check the paths a field can travel

The visible place where a field is collected is only the beginning. A value may be copied into a customer record, exported to a spreadsheet, read by a routing rule, or carried into a periodic report. Check the actual field identifier and any known aliases, not just its display label.

Build a compact dependency register covering these paths:

  • Entry points: Customer-submitted records, staff-created records, imports, and any other intake routes that collect the value.
  • Operational decisions: Assignment, qualification, onboarding, service handling, and exception queues.
  • Outputs: Reports, exports, document preparation, message preparation, and recurring management reviews.
  • Technical consumers: Validation rules, calculations, scheduled jobs, connected systems, and downstream copies.
  • Human workarounds: Personal spreadsheets, saved views, checklists, and instructions outside the main system.

For each suspected dependency, distinguish reference from reliance. A report containing the field does not establish that anyone acts on it. Conversely, a person who reads a weekly export may have a real dependency even if the source system shows no obvious activity.

Use available configuration reviews, process walkthroughs, and owner interviews together. If access to a consumer cannot be verified, mark it unknown. Do not convert a missing answer into a successful dependency check.

3. Define a pause that can produce an honest result

A collection pause means new records in the agreed scope stop acquiring the value through the routes being tested. It does not mean staff quietly copy old values, infer replacements, or ask the customer through another channel. Those actions would hide the very dependency you are trying to observe.

Choose a cohort narrow enough to recover if something fails, but broad enough to exercise the suspected uses. Pausing a field only for records that never reach onboarding will not test an onboarding dependency.

Keep three conditions explicit:

  1. Historical values stay untouched. They are outside this decision. Separate new, uncollected values from existing populated records in the review.
  2. Missing must remain distinguishable. Do not insert a plausible-looking default that allows a process to appear healthy while changing its meaning.
  3. The pause requires approval. Name the person authorised to start it, the person monitoring it, and the person authorised to restore collection.

Choose the observation period around process events, not convenience alone. A pause that ends before a monthly report runs cannot settle whether that report needs the field. A limited cohort reduces exposure, but it also limits what you can conclude.

4. Filled fictional example: the collection-pause plan

Fictional worked example: All roles, volumes, durations, thresholds, and outcomes below are hypothetical. They illustrate a recommended operating design, not an existing product control or a customer result.

Decision and scope

  • Candidate field: Department name on new business enquiry records.
  • Suspected problem: Staff believe it is unused, but onboarding and reporting owners have not confirmed that.
  • Decision question: Can new standard enquiries progress through assignment, onboarding handoff, and the monthly intake review without collecting department name?
  • Scope: All new standard enquiries entering through the reviewed website and staff-entry routes during a hypothetical five-week period.
  • Outside scope: Partner imports and specialist enterprise enquiries, whose intake dependencies have not yet been reviewed. Existing values remain unchanged.
  • Accountability: The operations lead approves the pause and final decision. The data steward maintains the register. The intake supervisor checks daily operations. The systems owner confirms readiness and can restore the approved collection path.

Dependency checks before starting

  • Assignment: The sales operations owner reviews the active assignment rules. Finding: territory and enquiry type determine ownership; department name is not referenced. Evidence retained: dated review of the relevant rule set.
  • Onboarding: The onboarding lead walks through the standard handoff checklist. Finding: department name is displayed, but the handoff requires a named contact and service scope instead. Evidence retained: annotated checklist and owner confirmation.
  • Monthly reporting: The analyst checks the intake report. Finding: department name appears in a detail export but is not a grouping or calculation input. Evidence retained: report definition and a scheduled review of the next report run.
  • Connected output: The systems owner reviews the known enquiry export. Finding: the field is included. Whether its recipient requires a value remains unresolved. Start condition: the recipient must confirm acceptable missing-value behaviour before the pause begins.
  • Staff workarounds: Intake staff review their local handoff sheets. Finding: one sheet includes the field, but its owner identifies no decision based on it. Evidence retained: owner response and revised handoff instructions.

Observation and recovery

  • Start gate: Every known consumer has an owner and a recorded result; the export recipient confirms that missing department names are accepted. The operations lead authorises the start.
  • Daily observation: The intake supervisor checks assignment failures, records held for missing information, export rejections, and requests to collect department name elsewhere.
  • Weekly observation: The data steward reviews cohort progression and exception tickets with sales and onboarding owners.
  • Event observation: The analyst checks the monthly report run and confirms whether any review decision was blocked or altered.
  • Immediate review trigger: Any confirmed export rejection attributable to the missing field, or any standard enquiry that cannot be assigned or handed over without it.
  • Restart authority: The operations lead may authorise a scoped restart after the systems owner and affected process owner verify the failure. A technical incident may justify restoring collection while the dependency is investigated.
  • Recovery boundary: Restore only the affected route or cohort where feasible. Do not automatically seek the value from every customer whose record was created during the pause.

5. Record exceptions as evidence, not votes

A complaint is a useful signal, but it is not yet a reason to restore universal collection. Capture the blocked task, affected record, date, consequence, and attempted alternative. Keep customer details in an appropriately restricted operational record rather than copying them into a broadly shared decision log.

Also capture whether someone solved the problem by collecting the same value elsewhere. A pause that merely moves collection into email or private spreadsheets has not demonstrated that the information is unnecessary.

In the fictional example, the exception log contains these entries:

  • Week two: a salesperson requests department name for account preparation. Evidence: one enquiry, no blocked task, and no decision identified that depended on the value. Disposition: preference recorded; collection remains paused.
  • Week three: an onboarding coordinator reports an incomplete handoff. Evidence: the missing item is actually a named implementation contact. Department name would not identify that person. Disposition: record as a separate process issue, not evidence to restore the field.
  • Week four: an export recipient asks where the column values went. Evidence: export accepted; recipient confirms the column was displayed but not used in processing. Disposition: document the confirmation and retain the report-run check.

A genuine exception would look different: a specific process cannot complete, the owner can explain why, and the problem is attributable to the absent value rather than a simultaneous change. It may justify a narrow collection requirement rather than restoring the field for everyone.

6. Make quiet failures visible

No complaints is weak evidence if nobody knows what to report. Tell affected teams what is paused, what symptoms matter, where to record them, and who can authorise recovery. Ask about delayed work and alternative collection, not just system errors.

Compare the paused cohort with a relevant recent baseline where one is available. Use measures close to the suspected dependency: assignment completion, handoff holds, export acceptance, and report completion. Broad outcomes such as revenue are harder to interpret and may move for unrelated reasons.

In the hypothetical plan, the team reviews all 74 new in-scope enquiries. All have an assignment outcome, and all enquiries reaching the onboarding stage have a recorded handoff outcome. The export and monthly report both complete without a field-related blockage. These are fictional observations, not expected performance targets.

Check exposure as well as outcomes. If no enquiry reached onboarding, there is no onboarding evidence. If a rare workflow did not occur, explicitly exclude it from the conclusion or extend observation. “Not tested” is more useful than a confident but unsupported pass.

Pause or reassess the test if another major process change makes attribution impossible. Otherwise, a new routing failure could be blamed on the missing field, or an unrelated improvement could hide its effects.

7. Close with a decision record, not an indefinite pilot

Set the decision meeting before starting. Its output should be a bounded conclusion supported by the checks and exceptions, rather than a general statement that the field seems unnecessary.

Filled hypothetical final decision record

  • Decision: Retire department-name collection for the reviewed standard-enquiry routes.
  • Evidence: Pre-pause dependency checks completed; 74 in-scope enquiries reviewed; no confirmed field-related assignment or handoff blockage; export acceptance and monthly report review completed; three exception reports investigated without establishing a collection need.
  • Limit: This decision does not cover partner imports or specialist enterprise enquiries. Their continued collection is pending a separate dependency review, not a permanent exemption.
  • Historical data: No decision made about retaining, changing, or deleting existing values.
  • Follow-through: The data steward updates the field-purpose register and relevant operating instructions. The operations lead confirms the retired collection does not reappear through an unreviewed intake route.
  • Reopening condition: A process owner presents a verified blocked task or a new approved purpose, identifies the minimum cohort needing the value, and explains why an available alternative is insufficient.
  • Follow-up: Review the next monthly reporting cycle and the outstanding intake routes. Owner: operations lead.

If evidence supports restoring collection, record the purpose, affected cohort, and observable need just as carefully. If evidence remains incomplete, extend the pause only to answer a named question with a scheduled observation. An ownerless extension simply recreates the original uncertainty.

8. Start with one field and one scheduled event

Your next action is to choose one questionable field and book a dependency review with its collection owner, a downstream process owner, and someone who can inspect its technical consumers. Bring the field identifier, intake routes, and one suspected use.

Leave that review with a named decision owner, a bounded cohort, an exception log, a restart condition, and a date tied to the next meaningful downstream event. If a critical dependency remains unknown, investigate it before starting. The useful result is a defensible collection decision—not simply fewer populated fields.

Data stewardshipCustomer dataCRM operationsProcess improvement

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