Profile missingness by context
Measure missing fields by time period, source system, organization, and workflow state. A column that is mostly complete overall may be absent for an entire business unit. Compare missingness with the task’s required inputs, rather than using one completeness percentage as a proxy for suitability.
Separate business meanings
An empty termination date can mean an active contract, an unavailable record, or a field that was never collected. Ask the source to distinguish these cases where possible. Keep zero quantities and empty strings separate from missing values until their meanings are established in the dictionary.
Inspect transformation effects
Exports, spreadsheet conversions, and joins can introduce gaps that were not present in the original system. Review unmatched keys, truncated files, and inconsistent category encodings. Preserve enough lineage to identify whether a missing value originated in the operational record or appeared during preparation.
Agree on treatment and limits
Specify which missing fields are acceptable, which records require exclusion, and which derived replacements need separate flags. Avoid silently filling unknown business facts with plausible values. Bring these rules to Fiund with sample rows and task requirements so prospective sources can be evaluated against a concrete acceptance plan. Review a small group of missing values with someone who understands the originating workflow. Ask what happened in the business process and what the export preserves. Keep the explanation with the affected field definition, since identical blank cells can require different treatment when they arise from different systems or process states.
Review checklist
- Profile gaps across meaningful slices
- Define blank, zero, and not-applicable values
- Trace gaps introduced by joins or exports
- Flag derived replacements explicitly