Lineage

See what a late table touches before you message anyone.

Tidewater reads your queries and your dbt project to draw what feeds what, from source tables to the dashboards people open. Late data stops being a mystery and becomes a list.

Request a demo How we build lineage from SQL

Built from your SQL

A graph that is never out of date.

Nobody draws the lineage by hand. Tidewater reads the queries your warehouse already ran and the models in your dbt manifest, so a new table appears in the graph the day it is first used.

Table and column lineage from query logs and dbt manifests
SELECT star, CTEs and renames resolved, not guessed
Sources, models, marts, dashboards and reports in one graph

Impact

From late table to affected people in one view.

Select a table and see everything downstream with its owner and how recently it was opened. The board pack that goes out at 08:00 shows up before you have to remember it exists.

Downstream assets ranked by how often they are opened
Owners on every node, so you know who to tell
The same list is attached to every alert

revenue_rollup

Mart, Refreshes at 07:00

finance-data

margin_by_sku

Mart, Refreshes at 07:00

finance-data

Exec revenue

Dashboard, Opened 14 times since 06:00

bi-team

Finance close

Dashboard, Opened 6 times since 06:00

finance-ops

Board pack

Report, Sent to 12 people at 08:00

finance-ops

Change safety

Know who reads a column before you drop it.

Column level lineage turns a schema change into a question with an answer. The diff shows what changed and which models read the column, so you can warn the owners first.

Readers of each column, up to the dashboards
Schema diffs linked to the models they affect
Available on the Business plan and above

discount_code

Added : varchar

shipping_cents

Type changed : integer to numeric(12,2)

legacy_region

Removed : was varchar

Read by 3 models

Under the hood

How it works, briefly.

Parsed, not sampled

Tidewater parses SQL statements into column references. It does not run your queries and it does not read your data.

Models and warehouse together

The dbt manifest names the models and owners. Query logs show what actually ran, including the ad hoc queries that models miss.

Fast on large warehouses

Incremental parsing keeps the graph current without reading the whole query history again. See the August changelog entry for the details.

See your own pipelines in Tidewater

Book a 30 minute walkthrough. We connect a sample of your setup and show you what it would have caught last month.

Request a demo Read the docs