Aglet

Verify a Missing-Field Data Quality Repair

Verification must show that the intended field is present for the right records without masking legitimate absence. Compare before and after results by source and record state, exercise the transformation with representative fixtures, and keep a check for new omissions. State what success cannot prove.

Check whether the outcome improved

  1. Define valid presence

    Write the field rule in observable terms: which record states require a value, which may remain empty, and what types or formats are accepted. Build examples for required, permitted-empty, malformed, and previously failing records so the check distinguishes repair from blanket defaulting.

  2. Compare affected cohorts

    Measure the missing rate before and after the change across the same source, time window, and processing version. Review preserved values, newly populated values, rejected records, and unrelated fields. Investigate any count movement that the proposed repair does not explain.

  3. Exercise forward protection

    Run contract and transformation checks with a complete payload, an omitted field, a null value, and a changed field type. Confirm the expected outcome for each case and add a bounded recurring review of missing rates or rejected inputs with an accountable owner.

What to carry forward

Accept the repair when required values meet the contract, legitimate empty states remain distinguishable, existing values are preserved, and forward checks fail clearly on a new omission. Record cohort limits and untested sources. If any result depends on an unverified assumption, keep the verification gate open.

Keep the decision with the work.

Use a Work Item in Aglet to record the problem, the evidence you have, and the next decision. Add an owner and priority, then keep updates in the discussion so the next person can follow the reasoning.

Create an account See the product workflow