Data Quality Vocabulary Exercises

1 / 5

A data quality dashboard shows a 2.3% null rate on a required field.

Which data quality dimension does this violate?

Data quality dimensions (DAMA framework):

Dimension	Question it answers	Example violation
Completeness	Are all required values present?	2.3% null rate on required field
Accuracy	Does data match reality?	Wrong address on customer record
Timeliness	Is data fresh enough for use?	Report uses yesterday's data
Consistency	Same values across systems?	CRM and warehouse disagree on revenue

Key vocabulary: null rate, completeness check, required vs. optional fields, data quality threshold.

2 / 5

A pipeline has a Great Expectations suite checking for data validity.

What does "data validity" mean as a quality dimension?

Validity = data meets defined business rules and constraints:

Validity check type	Example rule
Range validation	`age BETWEEN 0 AND 150`
Format check	Date must match `YYYY-MM-DD`
Enum constraint	status IN ('active', 'inactive', 'pending')
Regex pattern	Email must match `^[\w.]+@[\w.]+$`

Tools: Great Expectations expectations, dbt tests (accepted_values, not_null, relationships). Key vocabulary: constraint check, range validation, regex pattern check, referential integrity check.

3 / 5

A data monitoring alert fires: "Row count dropped from 1.2M to 890K yesterday."

In data monitoring, what does a sudden drop in row count typically signal?

Row count anomaly detection — a fundamental data monitoring check:

Anomaly type	Likely cause
Sudden drop in row count	Pipeline failure, source API issue, schema change filtering rows
Sudden spike in row count	Duplicate ingestion, re-processing of historical data
Gradual decline	Source system usage decline or data sourcing issue

Key vocabulary: data volume anomaly, ingestion failure, row count SLO, anomaly detection threshold, data observability (Monte Carlo, Anomalo).

4 / 5

An engineer says: "The data contract between the producer team and the consumer team was violated."

What is a data contract?

Data contracts = producer-consumer agreement on data structure and behaviour:

Contract element	What it defines
Schema	Field names, types, nullability
Freshness SLO	How recent the data must be
Volume expectations	Expected row count ranges

Prevents: "the upstream team changed a field name and broke my pipeline". Key vocabulary: schema contract, breaking schema change, producer obligations, consumer expectations, contract versioning.

5 / 5

A data reconciliation job reports: "0.4% discrepancy found between the source system and the warehouse."

What is data reconciliation?

Reconciliation = source vs. target comparison to verify complete, accurate ingestion:

Reconciliation check	What it verifies
Row count match	Same number of records source vs. warehouse
Aggregate match	sum(revenue) matches between systems
No dropped rows	No records lost or duplicated in transit

Discrepancy investigation: trace which pipeline stage introduced the difference. Key vocabulary: source-to-target reconciliation, reconciliation threshold (acceptable error %), discrepancy investigation, audit count check.