Ataccama Data Quality

Ataccama data testing

Ataccama checks SQL tables with reusable data quality rules that attach to catalog attributes. When a rule fails, Ataccama stores the invalid records so data stewards can review and fix them.

Compare: data testing tools

• SQL Table Tests: DQ rules run against catalog item attributes (SQL table columns) across completeness, uniqueness, validity (format/regex via conditions), accuracy (reference data lookups), and timeliness dimensions — plus custom rule conditions (docs).

• Inspect Failed Rows: The Invalid Samples feature lets users view actual records that failed DQ rules — selectable by attribute and rule, stored in object storage, with configurable retention. Must be enabled per catalog item (docs).

• DataFrame Tests: Ataccama ONE connects to data sources via SQL connectors; no in-memory Pandas, Spark, or PySpark DataFrame testing library documented (docs).

• Data Diff: No row-by-row comparison between two datasets documented in Ataccama ONE (docs).

• Data Contracts: No machine-readable producer-consumer schema and quality SLA agreements (YAML/JSON Schema data contracts) documented in Ataccama ONE (docs).

• Unstructured Data Tests: Integration with Snowflake Document AI for unstructured data validation is referenced externally but not found as a documented feature in the current ONE docs (docs).

• Table Comparison Tests: Accuracy rules check values against reference data, but explicit cross-table aggregate comparison (row count/sum matching between source and destination) is not documented as a named feature (docs).

Ataccama data observability

Ataccama profiles every column and detects anomalies in monitoring projects. Monitors are configured in the UI.

Compare: data profiling tools · data observability tools

• Data Profiling: Computes 55+ column-level statistics: null/not-null %, distinct and duplicate counts, min/max/mean/stddev, value distributions, frequency analysis, quantiles, pattern/mask analysis, and data type detection (docs).

• Anomaly Detection: Two models: Isolation Forest (time-independent, for static metrics) and a time-dependent model (for sequential pattern deviations). Five configurable sensitivity levels (Very Low to Very High). Applied at attribute level in monitoring projects and catalog items (docs).

• Freshness Monitors: Data Observability module includes freshness monitoring that verifies tables receive timely updates — natively supported for Snowflake, PostgreSQL, BigQuery, and Oracle (docs).

• Volume Monitors: Anomaly detection on record counts detects sudden drops or spikes in data volumes; tracked over time with the time-dependent model (docs).

• Schema Monitors: Structure checks in monitoring projects detect missing columns, new/missing tables, and data type modifications — alerting on structural changes (docs).

• Field Monitors: Anomaly detection tracks attribute-level metrics (mean, stddev, variance, null count, duplicate count, distinct count) as time series and flags deviations using statistical models (docs).

• Distribution Monitors: Profiling captures frequency analysis and value distributions; the time-dependent anomaly detection model tracks frequency patterns over time and flags distribution shifts (docs).

• Partition-Based Monitors: Monitoring projects on metastore sources offer a 'last partition analysis' option — scoping monitoring to the most recent partition rather than full-table scans (docs).

• UI-Based Monitors: All monitoring, DQ rule configuration, and anomaly detection settings are configurable through the Ataccama ONE web UI without writing code (docs).

• Window-Based Monitors: No tumbling or sliding time window evaluation for intra-day metric monitoring documented in Ataccama ONE (docs).

• Pipeline Monitoring: No pipeline job runtime monitoring (dbt run times, Airflow DAG durations) documented in Ataccama ONE (docs).

• Segment Monitors: Monitoring projects support 'data slices' to narrow monitoring scope, which may qualify as segment-level monitoring — but GROUP BY dimension-level monitor breakdowns are not explicitly documented (docs).

• Custom SQL Monitors: DQ rules support custom conditions and the AI can generate SQL expressions; a dedicated 'custom SQL monitor' that tracks a user-defined SQL expression as a metric over time is not explicitly documented (docs).

• Monitors-as-Code: GraphQL API can trigger DQ evaluations and manage entities programmatically, but YAML/Python code-first authoring of monitors with version control and CI/CD deployment is not documented (docs).

Ataccama incident management

Ataccama's public docs don't describe an incident workflow for triaging and assigning failures.

• Alert Notifications: Notifications delivered via Slack and Microsoft Teams when data quality or observability issues are detected; configurable notification templates (docs).

• Incident Triage: No structured incident triage workflow (classify, assign owners, set priority, group related failures) documented in Ataccama ONE (docs).

• Incident Resolution: No incident resolution lifecycle (mark as resolved, log root cause, track fix) documented in Ataccama ONE (docs).

• On-Call Integrations: No PagerDuty, Opsgenie, or VictorOps integration documented; only Slack and Microsoft Teams notification channels are confirmed (docs).

• Alert Feedback: Anomaly detection has configurable sensitivity levels (Very Low to Very High) but explicit false-positive marking or per-anomaly user feedback to retrain the model is not documented (docs).

• Ticketing Integrations: No Jira or ServiceNow ticket creation integration found in the Ataccama ONE docs navigation; notification channels documented are Slack and Microsoft Teams only (docs).

Ataccama data quality analytics

Ataccama scores data quality and shows it in dashboards.

• Data Quality Dashboards: Data Observability Dashboards provide quality KPI visualization, trend analysis, and monitoring summaries across data assets (docs).

• Data Health Scores: The Data Trust Index provides a real-time, per-asset quality signal combining DQ rule results and observability metrics into a single health indicator (website).

• Incident Resolution Metrics: Ataccama has issue/incident management workflow configuration, but no documented MTTD/MTTR dashboard, incident volume trend, or resolution-rate-by-owner analytics was found in docs (docs).

Ataccama data lineage

Ataccama maps lineage at table and column level, end to end, with impact analysis.

Compare: data lineage tools

• Table-Level Lineage: Catalog item-level lineage tracks table-to-table data flows across supported technologies (docs).

• Column-Level Lineage: Attribute-level lineage tracks field-to-field dependencies, with dynamic switching between catalog item and attribute granularity in the lineage diagram (docs).

• End-to-End Lineage: Lineage spans the full data stack: ingestion sources (S3, ADLS), ETL tools (dbt, ADF, Glue), warehouses (Snowflake, BigQuery, Databricks), and BI (Power BI, Tableau) (docs).

• Impact Analysis: 'Impact and lineage analysis' is listed as core lineage functionality — showing downstream assets affected by data quality issues or schema changes (docs).

Ataccama data governance

Ataccama's data catalog and data products make it a data governance platform.

Compare: data catalog tools · data stewardship tools

• Data Catalog: Searchable catalog with full-text search, metadata tags, ownership (stewardship roles), business glossary integration, data profiling, quality scores, lineage, and live data preview (docs).

• Data Products: Ataccama ONE has a dedicated "Governed data products" capability letting teams define business-meaningful data products organized within domains/business processes, with a Data Trust Index (quality score) surfaced at the data-product level (website).

• Data Catalog Integration: Ataccama ONE IS a data catalog; it does not push quality metadata into external catalog platforms (Atlan, Alation, Collibra) (docs).

• Data Classification: Data Protection Classification in Ataccama ONE is a manual governance tagging system — users manually apply tags to glossary terms that propagate to attributes. No automated scanning of column values or names for PII/PHI/PCI (docs).

Ataccama workflows

Ataccama schedules data quality checks, exposes REST and GraphQL APIs, and integrates with dbt. It has no Python SDK.

• Scheduling: Monitoring projects and DQ evaluations support automated scheduling for profiling runs, test runs, and monitor checks (docs).

• REST / GraphQL API: GraphQL API for querying and managing catalog items, terms, lineage, and DQ evaluations — authenticated via Basic or Bearer tokens. Separate Audit API also available (docs).

• MCP Server: MCP Server is listed as a core integration interface on the Ataccama ONE homepage, connecting AI agents to governed data sources (website).

• dbt Integration: dbt is listed as a supported lineage technology — Ataccama ONE imports dbt lineage metadata to build end-to-end data lineage (docs).

• Python SDK: No pip-installable Python SDK documented; programmatic access is via GraphQL API (docs).

• CI/CD Integrations: The GraphQL API can trigger DQ evaluations programmatically, enabling CI/CD pipeline integration, but dedicated CI/CD integration documentation (GitHub Actions, pipeline blocking on failure) was not found (docs).

• Webhooks: Notification templates are documented but whether HTTP webhook callbacks (distinct from Slack/Teams) are supported as a notification channel is not confirmed (docs).

Ataccama AI agents

Ataccama writes data quality rules from plain-language prompts and suggests rules automatically. An MCP server lets AI agents query governed data through Ataccama.

• Natural Language Test Generation: Ataccama's Generative AI includes a Text-to-DQ Rule feature letting users create and debug data quality rules from natural-language descriptions (documented example: "Create a DQ rule to validate email addresses") (docs).

• Automated Test Generation: AI-powered 'Rule Suggestions' automatically recommends DQ rules from the existing rule library based on data profiles; 'Find similar rules' avoids duplication — all without manual authoring (docs).

• AI Issue Triaging: No AI-powered root cause analysis of data quality incidents documented in Ataccama ONE; anomaly detection identifies issues but does not reason about causes (docs).

• AI Issue Resolution: No AI-suggested or automated remediation for data quality issues documented in Ataccama ONE (docs).

• AI Chat Assistant: Generative AI features are embedded in the ONE UI (generate rules, explain SQL, write descriptions) but whether these constitute a conversational in-product chat interface vs. discrete AI-assist buttons requires further verification (docs).

Ataccama master data management

Ataccama ONE MDM matches and merges records into golden records, with stewards approving the matches.

Compare: master data management tools