Analytics overview

Coalesce Quality analytics gives you a bird’s-eye view of the health of your data stack so you can systematically improve what matters most

Analytics shows the state of your data in one place, so everyone in your company works from the same numbers. Use it to:

• Benchmark and monitor the performance of your most important data assets
• Track data quality metrics and show stakeholders how they improve
• Find gaps by owner, platform and data product

Data quality dimensions

Analytics groups every test into one dimension, based on the test type. See How metrics are calculated for the full list of test types in each dimension.

Dimension	Description
Accuracy	Data correctly represents real-world facts (e.g., accepted_values and accepted_range tests, custom SQL checks for calculated metrics).
Completeness	All necessary data is present (e.g., not_null tests on critical columns, volume monitors).
Consistency	Related data agrees across tables (e.g., a relationships test between a foreign key and its parent table, reconciliations between a source and its replica).
Uniqueness	There are no duplicate entries (e.g., unique tests on primary key columns).
Timeliness	Data is fresh and updated on time (e.g., freshness monitors, dbt_utils.recency, and model, source and node runs).
Validity	Data follows formats and rules (e.g., dbt_utils.expression_is_true, regex-pattern and not_accepted_values audits, segmented monitors).
Uncategorized	Every other test

Data quality metrics

Each metric is calculated per week or per month, for each dimension. How metrics are calculated gives the exact formulas.

Metric	What it tells you
% of tests without issues	The overall health of your tests. A low value means you have systemic issues, or tests you should review.
Number of tests	How broad your test coverage is.
Number of tests with issues	Where the problems are. A high or growing count shows the areas to fix first.
New and resolved issues	How fast your team resolves issues. If new issues grow faster than resolved ones, data debt is growing.
MTTR (hr)	The mean time to resolve an issue, in hours.
# Incidents	How many tests were part of an incident you declared. Incidents mark the most severe issues.

Acting on data quality insights

• Use Add Filter to segment the insights by owner, data product or platform, so each team sees its own numbers.
• In a larger team, write a runbook for each data quality dimension. For example, if the Timeliness score is low, add a source freshness check in your transformation layer (dbt, Coalesce Transform) or an automated freshness monitor.