Data & Analytics

Data Quality Score DQS

Director Analytics CTO VP Engineering COO

Data Quality Score is a composite metric that assesses the overall quality of data across dimensions including accuracy (data is correct), completeness (no missing values), consistency (same data across systems), timeliness (data is current), and uniqueness (no duplicate records). Low data quality erodes trust in analytics, leads to poor decisions, and increases engineering time spent on data debugging rather than analysis.

Data quality should be measured at the pipeline, table, and field level to enable targeted remediation rather than just tracking an aggregate score.

Formula

Weighted composite of accuracy, completeness, consistency, timeliness, and uniqueness checks

Where It Lives

dbt testsAutomated data quality tests in transformation pipelines
Great ExpectationsData validation framework with quality score tracking
Monte CarloData observability and quality anomaly detection
AtlanData catalog with quality scores and lineage tracking

What Drives It

Source system data entry quality (garbage in, garbage out)
Pipeline transformation logic errors or edge case handling
Schema changes in source systems breaking downstream pipelines
Missing data from failed API calls or integration gaps
Duplicate record creation from system integration issues

Causal Analysis: Implementing automated data quality tests at each pipeline stage causally prevents quality issues from propagating downstream, and can be evaluated by tracking DQS before and after test implementation.

Benchmark

Data quality scores above 90% across all dimensions are generally considered excellent; below 80% typically signals systematic data collection or pipeline issues requiring urgent attention.

Common Mistake

Treating data quality as a one-time project rather than an ongoing operational process with automated monitoring, leading to quality degradation between audit cycles.

How different roles think about this metric

Each function reads DQS through a different lens and takes different actions when it changes.

Director Analytics

The Director of Analytics owns data quality as a foundational responsibility. Without high data quality, no analytics output can be trusted, making it the prerequisite for all other data work.

CTO

The CTO ensures data quality monitoring is built into the engineering culture and that quality checks are automated in the CI/CD pipeline for data products.

VP Engineering

VP Engineering implements data observability tooling and embeds quality validation into data pipeline deployment processes.

COO

The COO relies on data quality as the foundation for operational reporting and escalates data quality failures that cause incorrect operational decisions.

Common Questions About Data Quality Score

Click any question to expand the answer.

What are the five dimensions of data quality?

The standard five dimensions are: Accuracy (data values are correct and reflect reality), Completeness (required fields and records are present), Consistency (the same information appears consistently across systems and time periods), Timeliness (data is available within the expected freshness window), and Uniqueness (no unintended duplicate records exist). Some frameworks add a sixth: Validity (data conforms to defined formats and business rules). Each dimension requires different monitoring approaches.

What is a data quality SLA and how should one be set?

A data quality SLA defines the minimum acceptable quality threshold for specific datasets used in business-critical decisions. For example: revenue data must be 100% complete with zero duplicate transactions; marketing attribution data must be 90% tagged. SLAs should be defined collaboratively with business stakeholders based on the decision impact of quality failures. Violations should trigger automated alerts and SLA breach notifications to data consumers.

How do I implement automated data quality testing?

Use a framework like dbt (with its built-in test types: not_null, unique, accepted_values, relationships) or Great Expectations to define expected data characteristics as automated tests. Run these tests in your CI/CD pipeline after each transformation job. Tests should cover null checks, referential integrity, value range validation, row count thresholds, and freshness checks. Route test failures to PagerDuty or Slack alerts for on-call data engineers.

What is data lineage and how does it support data quality management?

Data lineage is the ability to track where data originated, how it was transformed, and where it flows downstream. When a data quality issue is detected in a dashboard, lineage tools let engineers trace the problem back to its source within minutes rather than hours. Tools like dbt's documentation, Atlan, and Monte Carlo provide lineage visualization. Without lineage, debugging data quality issues becomes exponentially more difficult as pipeline complexity grows.

Related Metrics

Metrics that are commonly analyzed alongside DQS.

Data Freshness / Latency | Data Freshness / Latency Attribution Model Coverage | Attribution Model Coverage Dashboard Adoption Rate | Dashboard Adoption Rate Error Rate | Error Rate

Role guides that include this metric

See how each role uses DQS in context with the full set of metrics they own.

Director Analytics Guide → CTO Guide → VP Engineering Guide → COO Guide →

free, one business day

Want to know what is holding your Data Quality Score back?

Give us your website and we will send back the specific pages hurting your Data Quality Score, in the order worth fixing. You will have it within one business day, free, with no obligation.