FIELD NOTE · Database Marketing
Complete data is not necessarily accurate data
A filled field can still be wrong. Keep those two questions separate.
Define completeness narrowly
Completeness asks whether a value is present where one is expected. If 400 customer records each require five fields, there are 2,000 expected field values. With 180 missing values, completeness is (2,000 − 180) ÷ 2,000, or 91%. The calculator uses that definition. It does not check whether the values describe real people or current businesses.
Accuracy needs another test
An email field containing “unknown” is filled but may not be useful. A correctly formatted address can belong to the wrong customer. Separate missingness, formatting validity, factual accuracy, and recency. Combining all of them into an unexplained score makes it harder to decide what to fix. A high completeness score cannot certify that a list is ready for a campaign.
Choose fields that matter to the decision
A shipping workflow may require a delivery address. An email campaign may need a usable email, a meaningful segment, and an appropriate permission basis. Do not count optional fields as missing simply to produce a bigger problem. Write down which fields are required, why, and when an exception is legitimate.
Report the denominator
Always report records examined, required fields, missing values, exclusions, and the date of extraction. If a second report uses a different set of fields, its percentage is not directly comparable. For recurring checks, preserve a definition document and investigate changes in the population before crediting a cleanup effort. Improvement should describe the same measure applied to a comparable dataset.
Prepared with AI assistance. Examples are constructed for explanation, not reported customer results. Corrections are welcome through the contact page.
← All guides