Analytics Data Quality Assessment

Description

Judge whether an analytics dataset is fit for use by counting records that fail completeness, consistency, validity, accuracy and timeliness, before drawing conclusions.

Data quality assessment asks whether data are fit for a specific use, not whether they are perfect. Five aspects are checked: completeness (required data present), consistency (no contradictions across sources, such as spelling variants), validity (conformance to rules, such as legal dates), accuracy (agreement with real-world facts) and timeliness (current and delivered on time). Methods include profiling for missing values, outliers and type mismatches; cleansing; validation against predefined rules; governance policies; and auditing against reality or regulation. Each aspect gets a simple count-based metric, such as the share of records missing, invalid or older than a threshold. Metrics are chosen per data type and goal and themselves reviewed, and decisions about exclusions are documented so later analysts can follow them.

Attributes

Loss type
Siloed
Effort
Days

Steps

  1. Define the intended use of the dataset and the questions it must answer.
  2. Profile the data for missing values, outliers, duplicates and type mismatches.
  3. Validate records against rules (allowed values, date ranges, event sequences).
  4. Count records failing each aspect: completeness, consistency, validity, accuracy and timeliness.
  5. Cleanse or flag failing records and document each exclusion decision.
  6. Audit a sample against reality, for example comparing tracked events with a manual test session.
  7. Repeat the assessment whenever data are refreshed or tracking changes.

Sources

  • Understanding Audiences via Analytics (Jansen et al., 2024)
  • Data Quality Assessment (Pipino, Lee & Wang, 2002)
  • Build a Career in Data Science (Nolis & Robinson, 2020)

Also known as

  • Data quality assessment for analytics datasets
  • Data quality audit
  • Five-dimension data quality check

Categories

  • Research Operations
  • Research Design & Rigor