Kestrel Credit is a fictional company used in this demonstration site. The story below shows the kind of result a finance reporting team can expect and is not a real customer account.
The problem
Kestrel’s finance team signs off a loan book report on the first working days of each month. The report depends on twelve source feeds, including payment processors, a collections platform and a servicing system, which arrive in BigQuery at different times and are combined by dbt Cloud jobs.
When one feed was late, the finance team found out by opening the report and noticing that a balance looked low. Finding the cause meant asking three teams. In a bad month the sign-off slipped from the first working day to the third.
What the team did
Instead of monitoring every table, they described the close as a short list of dependencies.
- A deadline for each feed. The twelve feeds each got an expected arrival time and an owner. A feed that is thirty minutes past its time raises an alert to the owner and to the finance systems channel.
- Lineage from the report backwards. Tidewater shows the Tableau workbook, the four dbt models it reads and the feeds behind them. When a feed is late, the alert says which part of the report will be wrong.
- A close dashboard. A shared view lists the twelve feeds with their state, so on the last day of the month everyone watches the same page.
The result
Over two quarters the report was signed off on the first working day in five of six months, against two of six before. Finance escalations about stale dashboards fell by 84 percent.
The less obvious gain was ownership. Three of the twelve feeds had no clear owner when the project began. Setting a deadline forced the question, and each of those feeds now has a named person and a documented fallback.
What they would tell another team
Write the deadline in terms the business understands: “the servicing extract by 05:00”. Then let the platform do the arithmetic about which tables and dashboards that implies.