# Snowflake Data Quality

> Native Snowflake data quality checks built on data metric functions (DMFs) and expectations, scheduled per table, with volume and freshness anomaly detection and AI-suggested checks from Cortex Data Quality.

Source: https://toolsfordata.com/tools/snowflake-data-quality/ · Updated 2026-11-17

[Docs](https://docs.snowflake.com/en/user-guide/data-quality-intro)

Best for: Snowflake teams on Enterprise Edition that want data quality checks without adding a vendor or moving data.

## Overview

- Core Strength: Snowflake-native checks — DMFs, anomaly detection, failed rows and lineage in Snowsight
- Test Types: Partial — SQL, table compare, failed rows
- Detection: Full — profiling, anomaly detection, pipeline monitoring
- Monitor Coverage: 10 of 11 types — as-code + UI
- Incident Workflow: Partial — ticketing, on-call
- Quality Analytics: Partial — dashboards
- Lineage Depth: Full — table, column, end-to-end, impact analysis
- Governance Overlap: Partial — catalog, catalog sync, classification
- FinOps & AI Observability: Partial — query performance, cost monitoring
- AI Agent Features: Partial — chat, auto test gen, AI triage
- Developer Surface: Partial — MCP, dbt, scheduling
- Deployment Model: Partial — SaaS, cloud-native app
- Security & Access: Full — SSO, SCIM, RBAC, audit log…
- Pricing: Usage — Snowflake credits, Enterprise Edition
- Free Option: None
- Notable Caveat: Snowflake only; Enterprise Edition; anomaly detection and dashboard in preview

## What it's used for

Snowflake Data Quality is used to check tables inside Snowflake without adding a vendor: you attach data metric functions to a table, set an expectation, and Snowflake runs them on a schedule and flags volume and freshness anomalies.

## Pricing

Snowflake Data Quality has no separate license. Data metric functions need Snowflake Enterprise Edition or higher and run on serverless compute billed in Snowflake credits.

## Deployment

Snowflake Data Quality runs inside your Snowflake account on serverless compute; there is nothing to install.

## Review

Strengths:
- No extra vendor: checks run inside Snowflake, so data never leaves the account.
- System data metric functions cover nulls, duplicates, freshness and row counts; custom ones are SQL.
- Volume and freshness anomaly detection, plus an account-wide monitoring dashboard.

Caveats:
- Needs Enterprise Edition or higher.
- Snowflake only: nothing outside the account is checked.
- Every scheduled check adds serverless credits to your bill.

Recommendation: On BigQuery or Databricks, use their native checks: BigQuery Data Quality or Databricks Data Quality. For checks across several platforms, consider Soda Cloud or Monte Carlo; if you already run dbt, dbt tests may be enough.

## Pricing

- Model: platform usage
- Source: https://docs.snowflake.com/en/user-guide/data-quality-intro (checked 2026-11-10)

## Features

### Data Testing

- SQL Table Tests: Yes — System data metric functions (DMFs) count nulls, duplicates, accepted values, ranges, referential integrity and freshness, and custom DMFs run any SQL rule. ([source](https://docs.snowflake.com/en/user-guide/data-quality-system-dmfs))
- DataFrame Tests: No — DMFs attach only to tables, views and other table-like objects in Snowflake, not to in-memory DataFrames. ([source](https://docs.snowflake.com/en/user-guide/data-quality-intro))
- Unstructured Data Tests: No — INVALID_JSON_COUNT checks only that JSON is valid; there is no JSON Schema or LLM-based check. ([source](https://docs.snowflake.com/en/user-guide/data-quality-system-dmfs))
- Table Comparison Tests: Yes — ROW_COUNT_DIFFERENCE and REFERENTIAL_INTEGRITY_COUNT compare two tables, and a custom DMF can take a second table to compare aggregates such as sums. ([source](https://docs.snowflake.com/en/user-guide/data-quality-system-dmfs))
- Data Diff: No — DMFs return aggregate values or counts of failing rows; there is no row-by-row comparison of two datasets. ([source](https://docs.snowflake.com/en/user-guide/data-quality-system-dmfs))
- Data Contracts: No — Checks are DMFs with expectations attached to tables; there is no producer-consumer contract. ([source](https://docs.snowflake.com/en/user-guide/data-quality-expectations))
- Inspect Failed Rows: Yes — SYSTEM$DATA_METRIC_SCAN and the View failed records button return the rows behind a failing system DMF. ([source](https://docs.snowflake.com/en/user-guide/data-quality-fixing))

### Data Observability

- Data Profiling: Yes — The data profile tab shows row count, last update, null counts, min and max, and most common values per column. ([source](https://docs.snowflake.com/en/user-guide/data-quality-profile))
- Anomaly Detection: Yes — Anomaly detection (preview) learns from DMF history and flags values outside the predicted range, for ROW_COUNT and FRESHNESS only. ([source](https://docs.snowflake.com/en/user-guide/data-quality-anomaly))
- Freshness Monitors: Yes — The FRESHNESS DMF measures time since the last update, with an expectation or anomaly detection. ([source](https://docs.snowflake.com/en/user-guide/data-quality-system-dmfs))
- Volume Monitors: Yes — ROW_COUNT and the ROWS_IN_PAST_* DMFs track row volume, and ROW_COUNT supports anomaly detection. ([source](https://docs.snowflake.com/en/user-guide/data-quality-system-dmfs))
- Schema Monitors: Yes — SCHEMA_CHANGE_ADD_COUNT, SCHEMA_CHANGE_DROP_COUNT and SCHEMA_CHANGE_TYPE_COUNT count column changes between runs. ([source](https://docs.snowflake.com/en/user-guide/data-quality-system-dmfs))
- Field Monitors: Yes — Column DMF results are stored and charted as run history in Snowsight, with fixed-threshold expectations; anomaly detection does not cover column DMFs. ([source](https://docs.snowflake.com/en/user-guide/data-quality-ui-monitor))
- Distribution Monitors: No — DMFs such as APPROX_QUANTILE return values, but no distribution drift detection is documented. ([source](https://docs.snowflake.com/en/user-guide/data-quality-expectations))
- Segment Monitors: Yes — WITHIN GROUP on a DMF association returns one result per value of a group-by column. ([source](https://docs.snowflake.com/en/user-guide/data-quality-group-by))
- Custom SQL Monitors: Yes — Custom DMFs are SQL functions that run on the table's schedule, with results stored and charted over time. ([source](https://docs.snowflake.com/en/user-guide/data-quality-custom-dmfs))
- Partition-Based Monitors: Yes — A FILTER clause on a DMF association limits the check to rows matching a predicate, such as recent dates. ([source](https://docs.snowflake.com/en/user-guide/data-quality-filter))
- Window-Based Monitors: Yes — ROWS_IN_PAST_HOUR, DAY, WEEK, MONTH and YEAR count rows in trailing windows at each run. ([source](https://docs.snowflake.com/en/user-guide/data-quality-system-dmfs))
- UI-Based Monitors: Yes — The Data Quality tab in Snowsight lets users add checks, accept suggested ones and set the schedule without SQL. ([source](https://docs.snowflake.com/en/user-guide/data-quality-ui-setup))
- Monitors-as-Code: Yes — Checks are SQL statements (ALTER TABLE ... ADD DATA METRIC FUNCTION ... EXPECTATION) that can live in version control. ([source](https://docs.snowflake.com/en/user-guide/data-quality-working))
- Pipeline Monitoring: Yes — Snowsight shows run history for Snowflake tasks and task graphs and can retry failed tasks. ([source](https://docs.snowflake.com/en/user-guide/ui-snowsight-tasks))

### Incident Management

- Alert Notifications: Yes — Sends email, or Slack and Teams messages through webhooks, when an expectation fails or an anomaly is found. ([source](https://docs.snowflake.com/en/user-guide/data-quality-notifications))
- Incident Triage: No — The dashboard groups open incidents by type, but there are no owners, priorities or assignment. ([source](https://docs.snowflake.com/en/user-guide/data-quality-centralized-dashboard))
- Alert Feedback: No — Users can change sensitivity or turn off anomaly detection, but cannot mark a point as expected or a false positive. ([source](https://docs.snowflake.com/en/user-guide/data-quality-anomaly))
- Incident Resolution: No — No resolve or close workflow for incidents is documented. ([source](https://docs.snowflake.com/en/user-guide/data-quality-centralized-dashboard))
- Ticketing Integrations: Yes — Webhook notification integrations can post to a Jira automation webhook or ServiceNow. ([source](https://docs.snowflake.com/en/user-guide/notifications/webhook-notifications))
- On-Call Integrations: Yes — Webhook notification integrations can trigger PagerDuty events. ([source](https://docs.snowflake.com/en/user-guide/notifications/webhook-notifications))

### Data Quality Analytics

- Data Quality Dashboards: Yes — An account-wide dashboard (preview) shows healthy tables, coverage and open incidents by type. ([source](https://docs.snowflake.com/en/user-guide/data-quality-centralized-dashboard))
- Data Health Scores: No — The dashboard shows the share of healthy tables, not a quality score per table. ([source](https://docs.snowflake.com/en/user-guide/data-quality-centralized-dashboard))
- Incident Resolution Metrics: No — No MTTD or MTTR metrics; the dashboard counts open incidents. ([source](https://docs.snowflake.com/en/user-guide/data-quality-centralized-dashboard))

### Data Lineage

- Table-Level Lineage: Yes — Snowflake lineage in Snowsight shows upstream and downstream tables and views, linked from data quality results. ([source](https://docs.snowflake.com/en/user-guide/ui-snowsight-lineage))
- Column-Level Lineage: Yes — Snowsight lineage traces upstream and downstream columns. ([source](https://docs.snowflake.com/en/user-guide/ui-snowsight-lineage))
- End-to-End Lineage: Yes — External lineage accepts OpenLineage events from tools such as dbt and Airflow; BI tools are not named. ([source](https://docs.snowflake.com/en/user-guide/external-lineage))
- Impact Analysis: Yes — The data quality side panel lists impacted downstream assets for a failing check. ([source](https://docs.snowflake.com/en/user-guide/data-quality-ui-monitor))

### Data Governance

- Data Catalog: Yes — Catalog Explorer in Snowsight browses databases, tables and views, each with a Data Quality tab. ([source](https://docs.snowflake.com/en/user-guide/data-quality-ui-monitor))
- Data Catalog Integration: Yes — Results show in Snowflake's own catalog (Catalog Explorer); no push to third-party catalogs is documented. ([source](https://docs.snowflake.com/en/user-guide/data-quality-ui-monitor))
- Data Classification: Yes — Sensitive data classification scans tables automatically and tags privacy categories such as email and name (Enterprise Edition). ([source](https://docs.snowflake.com/en/user-guide/classify-intro))
- Data Products: No — Monitoring is grouped by schema and table, not by data product. ([source](https://docs.snowflake.com/en/user-guide/data-quality-centralized-dashboard))

### FinOps

- Query Performance: Yes — Snowflake query history shows duration and bytes scanned per query, with a query profile for each step. ([source](https://docs.snowflake.com/en/user-guide/ui-snowsight-activity))
- Cost Monitoring: Yes — DATA_QUALITY_MONITORING_USAGE_HISTORY shows the credits used by DMFs. ([source](https://docs.snowflake.com/en/user-guide/data-quality-intro))

### LLMOps

- AI Observability: No — ML DMFs compare predicted and actual columns, but there is no monitoring of LLM or agent outputs. ([source](https://docs.snowflake.com/en/user-guide/data-quality-system-dmfs))

### Workflows

- Scheduling: Yes — DATA_METRIC_SCHEDULE runs a table's DMFs every N minutes, on a cron expression, or when the table changes. ([source](https://docs.snowflake.com/en/user-guide/data-quality-working))
- CI/CD Integrations: Unknown — SYSTEM$EVALUATE_DATA_QUALITY_EXPECTATIONS runs checks on demand, but no CI/CD integration or merge blocking is documented. ([source](https://docs.snowflake.com/en/user-guide/data-quality-expectations))
- Circuit Breakers: No — Expectations record violations in event tables for you to act on; no task, DML or pipeline is halted when a check fails. ([source](https://docs.snowflake.com/en/user-guide/data-quality-expectations))
- Webhooks: No — Webhook notifications are documented only for Slack, Teams, PagerDuty, Jira and ServiceNow, not generic endpoints. ([source](https://docs.snowflake.com/en/user-guide/notifications/webhook-notifications))
- REST / GraphQL API: Unknown — The Snowflake SQL API runs SQL statements over REST, but there is no data quality endpoint. ([source](https://docs.snowflake.com/en/developer-guide/sql-api/index))
- MCP Server: Yes — A Snowflake-managed MCP server lets AI agents run SQL and Cortex tools, which can query DMF results; it has no data quality tools of its own. ([source](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp))
- Python SDK: Unknown — The snowflake.core Python API does not cover DMFs; checks can only be run as SQL through the Python connector. ([source](https://docs.snowflake.com/en/developer-guide/snowflake-python-api/snowflake-python-overview))
- dbt Integration: Yes — dbt Projects on Snowflake runs dbt build and dbt test inside Snowflake, and dbt can send lineage to Snowflake through OpenLineage. ([source](https://docs.snowflake.com/en/user-guide/data-engineering/dbt-projects-on-snowflake))

### AI Agents

- AI Chat Assistant: Yes — Cortex Code opens in Snowsight from the data quality dashboard to investigate incidents in a chat (preview). ([source](https://docs.snowflake.com/en/user-guide/data-quality-centralized-dashboard))
- Natural Language Test Generation: No — Suggested checks come from metadata; there is no plain-English check builder. ([source](https://docs.snowflake.com/en/user-guide/data-quality-ui-setup))
- Automated Test Generation: Yes — Cortex Data Quality suggests checks from table metadata and usage, which users accept or edit (preview). ([source](https://docs.snowflake.com/en/user-guide/data-quality-ui-setup))
- AI Issue Triaging: Yes — The dashboard's Investigate action opens Cortex Code with the incident loaded for root-cause analysis (preview). ([source](https://docs.snowflake.com/en/user-guide/data-quality-centralized-dashboard))
- AI Issue Resolution: No — Remediation is documented as manual SQL on the failed rows; no AI-proposed fix is documented. ([source](https://docs.snowflake.com/en/user-guide/data-quality-fixing))

### Security

- SSO: Yes — Snowflake supports SAML 2.0 and OIDC single sign-on with Okta, Entra ID and other providers. ([source](https://docs.snowflake.com/en/user-guide/admin-security-fed-auth-overview))
- SCIM Provisioning: Yes — SCIM 2.0 provisions users and roles from Okta or Microsoft Entra ID. ([source](https://docs.snowflake.com/en/user-guide/scim-intro))
- RBAC: Yes — Snowflake roles and privileges control who can attach DMFs and view results, plus data quality viewer and admin roles. ([source](https://docs.snowflake.com/en/user-guide/data-quality-access-control))
- Audit Log: Yes — The ACCESS_HISTORY view logs who read and changed which objects, for compliance auditing (Enterprise Edition). ([source](https://docs.snowflake.com/en/user-guide/access-history))
- Compliance Certifications: Yes — Snowflake holds SOC 2 Type II and ISO 27001, among others. ([source](https://docs.snowflake.com/en/user-guide/intro-compliance))

### Deployment

- SaaS Deployment: Yes — Data quality is a managed Snowflake feature on serverless compute. ([source](https://docs.snowflake.com/en/user-guide/data-quality-intro))
- Hybrid Deployment: No — Everything runs inside Snowflake; there is no customer-hosted data plane. ([source](https://docs.snowflake.com/en/user-guide/data-quality-intro))
- Self-Hosted Deployment: No — Snowflake is a managed service only. ([source](https://docs.snowflake.com/en/user-guide/data-quality-intro))
- Cloud-Native App: Yes — DMFs run inside the Snowflake account on serverless compute; nothing is installed. ([source](https://docs.snowflake.com/en/user-guide/data-quality-intro))

### Open Source

- Open Source: No — A managed Snowflake feature; nothing is open source. ([source](https://docs.snowflake.com/en/user-guide/data-quality-intro))

## Integrations

5 integrations, each sourced from the vendor docs.

- Data Warehouse: Snowflake
- Notifications: Email, Slack, Webhook
- Open Table Formats: Apache Iceberg

## Alternatives

- [Databricks Data Quality](https://toolsfordata.com/tools/databricks-data-quality/): The Databricks equivalent, if your pipelines and tables live on Databricks.
- [BigQuery Data Quality](https://toolsfordata.com/tools/bigquery-data-quality/): The Google Cloud equivalent for BigQuery tables.
- [AWS Glue Data Quality](https://toolsfordata.com/tools/aws-glue-data-quality/): The AWS equivalent, paid per use inside Glue.
- [Soda Cloud](https://toolsfordata.com/tools/soda-cloud/): Consider Soda Cloud for data contracts that cover Snowflake and the rest of your stack, with AI agents that explain issues.
- [Monte Carlo](https://toolsfordata.com/tools/monte-carlo/): Consider Monte Carlo for ML monitors and lineage across Snowflake, BI and pipelines.
- [dbt tests](https://toolsfordata.com/tools/dbt-tests/): The free option if you transform with dbt: tests live next to the models.
