Data Quality Starter Kit

Lightweight, copy-paste checks for catching broken pipelines before your stakeholders do.

Planned

Motivation

Data quality problems are rarely exotic. The overwhelming majority are the same five failures — stale data, volume collapse, schema drift, null explosions, and duplicate rows — arriving in slightly different costumes.

Full observability platforms are great, but they’re heavy for small teams, and the first 80% of protection is available to anyone with a query runner and fifteen minutes. That’s the gap: a starter kit that assumes no platform, just queries you paste into whatever scheduler you already have.

Scope

A deliberately small kit — a SQL-first checklist covering the five common failure modes:

Freshness — is the table actually updating on schedule?
Volume — did row counts collapse or explode versus the trailing baseline?
Schema — did columns vanish, get renamed, or silently change type?
Nulls — did a key column’s null rate jump?
Duplicates — did grain break without telling anyone?

Each check ships as a plain query, an expected-output convention, and a one-line explanation of what it catches. Plus a short guide to wiring them into cron, GitHub Actions, or a dbt test — wherever you already live.

Status & roadmap

Freshness + volume checks (first)
Schema, null, and duplicate checks
Scheduler integration guide (cron / Actions / dbt)
A worked “this would have caught it” incident example

What success looks like

A stakeholder says “is the data fresh?” and someone on the team can answer with evidence in under a minute — because the checks were pasted in months ago and everyone forgot they were there. That’s the good outcome: boring, invisible reliability.