New: Cookbooks and AI ExplanationsStep-by-Step recipes to solve problems connected to Roadmaps and Cheat Sheets. Need more details? Use AI buttons for structured and simple explanations with concrete examples throughout the whole platform.Take a look
Every failing check blocks the deploy, so the team starts merging around the pipeline instead of through it.
What you'll have at the end
A short policy naming which checks are hard blockers and which are warnings, backed by real historical thresholds
You need
A pipeline where every data check already runs on each deploy, every failure currently blocks the same way, and you can pull each check's pass or fail history from wherever the results live.
Not covered
Writing a brand-new check for something nobody watches yet, or reworking one already-noisy check's own numeric range: this assumes the checks already exist and only need sorting.
Your loudest data quality alert stopped meaning anything
start there instead if it's one specific alert that's noisy, not a whole pipeline's severity policy.
Add a range check to a business metric
go there first if a check's own numeric range was never set from real data; this recipe assumes the checks already exist.
Draft a data contract that survives a schema change
reach for this instead when the goal is stopping a schema break at its source, before any check even runs.
Checked 26 Aug 2026
Part of the Data Observability cookbook