The Lineage Depth Matrix, summarized
Filled square documented, half square partial, open square none found on 10 Nov 2026. The full matrix carries every cell's evidence.
Sixteen lineage tools by depth, sorted by the Upstream total (0 to 100). Evidence read 10 Nov 2026.| Tool | Table | Column | Transformation logic | BI field | Metric | Systems named | Total |
|---|
| Euno | documented | documented | partial | documented | partial | 28 | 67.0 |
| DataHub | documented | documented | documented | documented | none found | 11 | 62.5 |
| Secoda | documented | documented | partial | partial | partial | 11 | 61.5 |
| Atlan | documented | documented | documented | partial | none found | 16 | 58.5 |
| Foundational | documented | documented | documented | documented | none found | 14 | 58.0 |
| Coalesce Catalog | documented | documented | partial | partial | partial | 12 | 57.0 |
| Collibra | documented | documented | documented | partial | partial | 20 | 56.5 |
| Alation | documented | documented | partial | partial | none found | 16 | 55.0 |
| dbt | documented | documented | documented | partial | partial | 3 | 54.0 |
| Snowflake | documented | documented | partial | partial | partial | 7 | 51.5 |
| SYNQ | documented | documented | documented | partial | none found | 15 | 50.5 |
| Monte Carlo | documented | documented | documented | partial | none found | 18 | 50.0 |
| Databricks Unity Catalog | documented | documented | none found | partial | none found | 6 | 47.5 |
| Select Star | documented | documented | partial | documented | partial | 22 | 46.5 |
| IBM Manta | documented | documented | documented | partial | none found | 2 | 40.5 |
| OpenLineage and Marquez | documented | documented | partial | none found | none found | 12 | 36.0 |
The matrix counts depths documented in full. The Depth of lineage score (weight 25) reads the same evidence and also credits partial depths: Euno documents BI-field lineage and partial transformation and metric lineage, scored 7 of 10, while DataHub and Foundational document four depths in full and no metric lineage was found for either, scored 6.
The top four, as scored
Rank 1
67.0
3 of 5 depthsagents 7/10
Column-level lineage stitched across warehouse, transformation, BI and security tools, with labels and persona-scoped context served to agents over MCP, SDK and API. Deepest documented level: BI field.
Read the profileRank 2
62.5
4 of 5 depthsagents 8/10
Open-source column-level lineage from a SQL parser, with documentation propagation on by default in OSS and an MCP server with per-user tokens. Deepest documented level: BI field.
Read the profileRank 3
61.5
2 of 5 depthsagents 7/10
Secoda automates table and column lineage from SQL and dbt and exposes it through an MCP server that respects team permissions. Deepest documented level: column.
Read the profileRank 4
58.5
3 of 5 depthsagents 7/10
Automatic column-level lineage from SQL parsing plus a Metadata Propagator app that pushes owners, certificates and tags downstream, and a hosted per-user-permissioned MCP server with 41 tools. Deepest documented level: transformation logic.
Read the profile
Where Euno is behind
Euno is cloud SaaS only and publishes no dollar prices; a team that needs lineage inside its own network, or a full catalog around it, will pair it with another tool.
Where each is strongest
- Euno: Depth of lineage, 7 of 10. Column-level lineage stitched across warehouse, transformation, BI and security tools, with labels and persona-scoped context served to agents over MCP, SDK and API.
- DataHub: Served to agents, 8 of 10. Open-source column-level lineage from a SQL parser, with documentation propagation on by default in OSS and an MCP server with per-user tokens.
- Secoda: Served to agents, 7 of 10. Secoda automates table and column lineage from SQL and dbt and exposes it through an MCP server that respects team permissions.
- Atlan: Served to agents, 7 of 10. Automatic column-level lineage from SQL parsing plus a Metadata Propagator app that pushes owners, certificates and tags downstream, and a hosted per-user-permissioned MCP server with 41 tools.
- Foundational: Coverage of the stack, 7 of 10. Foundational derives column lineage by analyzing source code (SQL via SQLGlot, Spark, ORMs, COBOL) across repositories, warehouses and BI.
- Coalesce Catalog: Price transparency, 7 of 10. Catalog builds column lineage from observed SQL and maps it to Looker, Tableau and Power BI fields, with an OAuth MCP server scoped to the user's role.
- Collibra: Served to agents, 8 of 10. Technical lineage via Edge with in-line code context across about 40 sources, and an MCP server with lineage, glossary and measure-tracing tools.
- Alation: Served to agents, 8 of 10. Lineage nodes inherit ownership, trust flags, quality scores and definitions from the catalog, with version-aware upkeep and a remote MCP server that hides tools the user cannot use.
- dbt: Price transparency, 7 of 10. Column lineage computed from the project's own SQL, with transformation formulas, description inheritance and refresh on every production run.
- Snowflake: Lineage that carries meaning, 6 of 10. Native column-level lineage mined from query history inside Snowflake, queryable by GET_LINEAGE and folded with tags and masking policies.
Picks by situation
A data lineage tool records what feeds what: which source tables a model reads, which columns a transformation turns into which output columns, and which dashboards and metrics sit at the end of the chain. Most build the map by parsing SQL and query logs, some from run events emitted by the pipeline (the OpenLineage model), and a few from the BI tool's own field definitions.
The tools in this comparison all document table and column lineage. They part ways after that: at the transformation logic inside a model, at the BI field a chart is built on, and at the metric a team reports. They also differ on whether the lineage carries the owner, the definition and the classification along the path, and on whether an AI agent can query it, scoped to the person it acts for.
Upstream scores those differences on seven criteria with weights published before any score. Catalogs, context platforms and MCP server directories are adjacent subjects; this site covers lineage.
Guides and glossary