Bigeye Data Observability

Bigeye data testing

Bigeye runs rule-based checks on SQL tables, compares tables across sources, and shows the rows that fail a check.

• SQL Table Tests: Custom SQL rules cover any assertion; built-in templates include null rates (completeness), uniqueness, value ranges (accuracy), format checks (validity), and referential integrity via SQL subquery (consistency) (docs).

• Table Comparison Tests: Bigeye Deltas compare similar datasets (dev vs prod, source vs destination) using metric-based comparison to detect aggregate-level drifts and validate replication (docs).

• Inspect Failed Rows: Debug Query shows a preview of rows affected by a failing check and provides an exportable query for full investigation in the data source; currently available for Identification Formats, Financial, Contact Information, and Uniqueness metric categories (docs).

• DataFrame Tests: Bigeye connects to SQL data sources via agents; DataFrame validation (Pandas, Spark, Polars) is not a supported test type (docs).

• Unstructured Data Tests: The Data Sensitivity module only scans structured data warehouse columns using classifiers for PII/PHI/PCI detection; LLM-based semantic validation of free-text column content is not supported (docs).

• Data Diff: Bigeye Deltas compare datasets at metric level (aggregates, distributions) — not row by row — so individual inserted, deleted, or changed records are not surfaced (docs).

• Data Contracts: Bigeye has no data contract feature; BigConfig provides YAML monitors-as-code but does not formalize producer-consumer SLAs in a machine-readable contract format (docs).

Bigeye data observability

Bigeye monitors freshness, volume, schema, fields, distributions, segments and custom SQL. You configure monitors in the UI or as code, and Bigeye also monitors pipelines.

Compare: data observability tools · data profiling tools

• Data Profiling: On-demand profiler computes column-level null %, distinct %, duplicates, data types, pattern recognition, and distributions — results feed autometric suggestions (docs).

• Anomaly Detection: Autothresholds dynamically learn from historical metric values and flag deviations without requiring users to set fixed bounds — applied automatically on deployment (docs).

• Freshness Monitors: Freshness metrics check table last-update timestamps via system metadata tables with autothresholds that learn load patterns; data-based fallbacks available for all sources (docs).

• Volume Monitors: Volume metrics track row count additions with autothresholds; Row Count metrics support hourly or daily bucketing for granular intra-day volume monitoring (docs).

• Schema Monitors: Schema change detection tracks column additions, removals, and type changes across all connected sources and fires alerts on structural changes (docs).

• Field Monitors: Column-level metrics (null %, distinct count, distributions, validity checks) are tracked as continuous time series with autothreshold alerting (docs).

• Distribution Monitors: The Distributions metric category tracks min, max, average, variance, skew, and kurtosis per column over time with autothreshold anomaly detection (docs).

• Segment Monitors: Grouped Metrics apply a SQL GROUP BY to a dimension column (e.g., region, product_type), generating individual metrics with separate autothresholds per group (docs).

• Custom SQL Monitors: Custom SQL rules let users define arbitrary SQL expressions as continuously tracked metrics with autothreshold alerting over time (docs).

• Partition-Based Monitors: Bigeye's Data Time Window metric type scans only a configurable time-based partition (e.g., last 24h) rather than the full table, reducing compute cost on large time-partitioned datasets (docs).

• Window-Based Monitors: Metric Window Types include Data Time (max row timestamp lookback) and Data Time Window (hourly/daily bucketed aggregation) for granular time-windowed evaluation (docs).

• UI-Based Monitors: Bigeye provides a no-code UI for deploying metrics, configuring thresholds, and managing monitors without requiring YAML, CLI, or code (docs).

• Monitors-as-Code: BigConfig is a YAML-based declarative monitors-as-code solution deployed via `bigeye-cli plan` and `apply` commands, enabling version-controlled metric management (docs).

• Pipeline Monitoring: Bigeye tracks pipeline reliability (freshness, volume) and integrates with dbt to surface job run status alongside data quality metrics for end-to-end pipeline health (docs).

Bigeye incident management

Bigeye triages issues, takes feedback on alerts, and opens tickets in Jira or ServiceNow.

• Alert Notifications: Bigeye sends metric failure alerts via Slack, email, Microsoft Teams, and webhooks, configurable per metric or collection (docs).

• Incident Triage: The Unified Issue Inbox provides structured workflows to classify, assign, prioritize, and group incidents; status syncs bidirectionally with Jira and ServiceNow (docs).

• Alert Feedback: Bigeye has a Feedback feature where users rate anomalies (thumbs up/down), mute alerts, and mark false positives — feeding back into the autothreshold model (docs).

• Incident Resolution: Bigeye tracks issue status through Triage→Acknowledged→Monitoring→Closed; bigAI provides suggested resolutions and preventions for root-cause investigation (docs).

• Ticketing Integrations: Bigeye has bidirectional Jira and ServiceNow integrations that create and sync tickets from data quality incidents — status and comments stay in sync across both systems (docs).

• On-Call Integrations: Bigeye notification channels include Slack, Microsoft Teams, Email, Jira, ServiceNow, and Webhooks; no PagerDuty, Opsgenie, or VictorOps on-call paging integration is documented (docs).

Bigeye data quality analytics

Bigeye's dashboards, health scores and resolution metrics track data quality over time.

• Data Quality Dashboards: Bigeye's dashboard and scorecard display monitoring coverage, health status per schema/table/column, and daily trend history across all connected sources (docs).

• Data Health Scores: The Scorecard shows the percentage of healthy metrics per data dimension at schema, table, and column level with color-coded (green/yellow/red) daily history (docs).

• Incident Resolution Metrics: The Bigeye Dashboard's 'Issue Response' section explicitly tracks average time to resolution (weekly trend), issues closed per user, and percentage of issues interacted with — operational response metrics, not just data health (docs).

Bigeye data lineage

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

Compare: data lineage tools

• Table-Level Lineage: Table-level lineage is tracked via dedicated lineage connectors for Snowflake, Databricks, SQL Server, Oracle, dbt, ADF, and BI tools (docs).

• Column-Level Lineage: Lineage Plus connectors parse lineage at the column level with no manual mappings, across databases, ETL platforms, and BI tools; table nodes expand to column lineage on demand (docs).

• End-to-End Lineage: Lineage connectors span ETL tools (dbt, ADF, Informatica, Matillion), SQL databases, and BI tools (Tableau, Looker, PowerBI, Qlik) for full-stack lineage (docs).

• Impact Analysis: From any incident, Bigeye automatically shows affected downstream tables and reports (Tableau, PowerBI, Looker) and their owners for blast-radius assessment (docs).

Bigeye data governance

Bigeye's catalog, data classification and data products overlap with data governance tools.

Compare: data classification tools · data catalog tools

• Data Catalog: The Bigeye Catalog provides a full inventory of connected data assets with ownership tracking, custom attributes, tags, global search, and hierarchical navigation (source → schema → table → column) — functioning as a governance-focused catalog alongside observability features (docs).

• Data Catalog Integration: Bigeye integrates with Alation, Atlan, and data.world to push data health and quality insights into external catalog solutions (docs).

• Data Classification: Data Sensitivity module scans sources with out-of-the-box PII, PHI, PCI, and financial classifiers built from regex, checksum, and ML detectors, plus custom classifiers, mapping each data class to a sensitivity level (docs).

• Data Products: Bigeye 'Data Domains' let you group tables, reports, and other assets by business function, team, or product line (assets can belong to one or multiple domains), and Collections similarly bundle monitors for a shared owned view (website).

Bigeye LLMOps

Bigeye observes AI agents and governs which data they can access.

• AI Observability: Agent Trust discovers AI agents, records their conversations and the tables they access, and connects to AI platforms such as Snowflake Intelligence, Databricks Genie, Claude Code and Cursor (docs).

Bigeye workflows

Bigeye has a REST API, a Python SDK, webhooks and a dbt integration, so you can manage monitors from code and pipelines.

• Scheduling: Metrics run on configurable cron schedules; bigAI provides cron suggestions and hourly/daily metric window types are supported (docs).

• Webhooks: Bigeye webhooks send JSON payloads to arbitrary URLs when metrics enter alerting state, configurable per metric or collection with custom headers (docs).

• REST / GraphQL API: Bigeye exposes a full REST API for programmatically triggering metric runs, retrieving results, managing data sources, and lineage operations (docs).

• MCP Server: Bigeye has a Docker-based MCP server enabling AI agents (Claude Desktop) to query data quality issues, lineage, and catalog via the Model Context Protocol (docs).

• Python SDK: The Bigeye Python SDK (pip-installable) provides programmatic metric management, data source operations, custom rule execution, and column-level lineage creation via DatawatchClient (docs).

• dbt Integration: Bigeye integrates with dbt Cloud (via API) and dbt Core (via CLI/Airflow) to link lineage, track job run status, and correlate data quality issues with dbt runs (docs).

• CI/CD Integrations: BigConfig CLI supports plan/apply workflows for monitors-as-code and Deltas validates dbt model changes, but blocking PR merges on quality failure is not documented (docs).

Bigeye AI agents

Bigeye writes checks from plain-language prompts, suggests checks automatically, and uses AI to triage issues and propose fixes. It also observes AI agents and ships an MCP server.

• Natural Language Test Generation: bigAI Chat lets users create metrics/data quality checks and custom SQL rules conversationally in plain English, with no query language required (docs).

• Automated Test Generation: Autometrics automatically recommend which metrics to deploy based on column type, semantic details, and data profiling results — available without manual configuration (docs).

• AI Issue Triaging: bigAI generates natural-language descriptions of data quality incidents and provides suggested resolutions to speed up root-cause analysis (private preview) (docs).

• AI Issue Resolution: bigAI 'Suggested Resolutions' analyzes issue metadata, underlying data patterns, and upstream ETL code to generate a list of potential fixes to resolve the problem — meeting the 'proposes a remediation action' criterion (docs).

• AI Chat Assistant: Bigeye has no in-product AI chat assistant; bigAI is an embedded suite of automated AI features (incident summaries, scheduling suggestions, impact assessment) rather than a conversational interface (docs).