Trace Every Signal to Its Source with Data Lineage in Qualytics

Picture the workflow you already know: an anomaly fires on a reporting table feeding an executive dashboard. You know something is wrong downstream, but you don't yet know where it started. So you leave the quality context you're working in, open a separate lineage tool, trace the table back through its sources, then return to figure out which of those upstream assets carry active issues. The investigation lives in two places, and the map you're reading has no idea which nodes are actually broken right now.

Data lineage is now available in Qualytics, and it's free for all customers through at least Q3. You get a visual graph of how data flows across your datastores, containers, and fields, with live anomaly counts woven directly into that graph. 

Three workflows it supports

Root cause investigation: When an anomaly fires on a reporting table, open lineage and walk the graph upstream. Every intermediate node shows its live anomaly badge, so you can see whether the problem is isolated or whether the same contamination exists at multiple stages. Click a badge and the side panel opens pre-filtered to that scope. You move from "something is wrong downstream" to "here is where it started" without leaving the quality context you're already working in.

Impact analysis before changes: Before deprecating a column, renaming a table, or restructuring a pipeline, use the downstream view to see every container and field that would be affected. Relationships can be added or removed inline as part of that review, giving your governance team a documented record of what changed and why.

Critical asset tracing: Trace any KPI, metric, or regulated field from its source across every container and computed asset where it appears, including across datastores. Whether you're tracking a number that feeds an executive dashboard, an input to an agent, or a PII field under BCBS 239, DORA, or SOX, you get a clear auditable path from source to consumption for every critical attribute in your environment.

How it works

The lineage graph lives on a dedicated tab on every container. It operates at two levels simultaneously. At the container level, you see how tables and files connect across your entire stack, including across bronze, silver, and gold layers of your data platform. At the field level, you can drill into any container node and see exactly which columns flow where, which fields carry active quality issues, and what their anomaly counts are right now.

Connections in the graph come from three sources, and they coexist without conflict. First, relationships you or your team draw manually in the UI. Second, lineage imported automatically from connected data catalogs: Atlan, Alation, Collibra, Purview, and DataHub. Third, edges that Qualytics generates automatically when you create or update computed containers. All three types live in the same graph. A container can have all three at once, and none of them overwrites the others.

Field-level lineage and why it matters

Most lineage tools reach the field level by parsing SQL query logs. They can map column-to-column relationships for what runs through the warehouse. That covers a lot of ground, but it has a structural gap: anything that doesn't emit a SQL log doesn't appear. Python transformations, stored procedures, external feeds, and API ingestion all move data in ways that SQL parsing cannot see. For those flows, the graph goes dark exactly where you may need it most.

In Qualytics, field-to-field connections can be drawn manually in the UI between any two fields in any two datastores, regardless of system type or how the data actually moves. If a Python job picks up a field from Postgres and lands it in Snowflake, you can map that connection directly. If a regulated field moves through a system that doesn't emit query logs, you document it here. These manually-created connections are first-class edges in the same graph alongside everything that comes from catalog integrations and computed containers. The result is a field-level map that reflects what your team actually knows about the data, not just what the warehouse can infer from its own logs.

For data catalogs like Atlan and Alation, lineage is a core strength, and if you have a catalog integration configured in Qualytics, those relationships already import automatically. The distinction is where the map sits. Catalog lineage lives in the catalog, separate from where validation runs. In Qualytics, the same graph that shows you the data flow is the surface where your quality checks govern those assets. The anomaly counts on each node aren't notifications from another system. They're the checks that are already running.

What's coming

Qualytics will add native lineage ingestion directly from Snowflake and Databricks in a future release. That removes the dependency on a separate catalog for lineage entirely, letting the platform itself serve as the lineage source without an additional integration layer.

Lineage is available now, free for all customers through August 2026. The tab is off by default and takes a few minutes to enable. Current Qualytics customers can reach out to their FDE to turn it on.