WNL Analytics · for SQL analysts new to dbt

You already write the SELECT.

dbt handles everything around it.

Interactive courses for analysts joining the dbt-analytics project — small steps, real commands, a certificate at the end of each. Plus a handbook for looking things up later.

The courses · in order

0Git essentials6 lessons · ~1 hrVersion control from zero — the foundation everything else stands onStart here if you have never used git. No installation needed yet — this course is about the ideas and the handful of commands.Open →1Understanding dbt9 lessons · ~1¼ hrsWhat dbt is and why it exists — in pictures, before you ever run itFor SQL analysts who have never used dbt. Nothing to install, nothing to run — short visual chunks with a question at each step. Take it after Git essentials, before Your first PR.Open →2Your first PR9 lessons · ~3.5 hrs hands-onA guided mechanics lab: from a blank machine to a merged staging modelHands-on, at your own machine. Assumes Git essentials and Understanding dbt (or equivalent experience). This course deliberately uses one missing staging model to teach the mechanics cleanly; most later work will discover, change and compose existing domain models. Best done with real access in your first week.Open →

Start with Git essentials if version control is new — everything else assumes it. Already fluent in git? Start with Understanding dbt, then use Your first PR as the practical capstone. Progress and certificates are saved in your browser.

Why dbt

Reused

Hundreds of tested models already exist. Most analyses start from another model, not a blank worksheet.

Ordered

dbt reads your SQL and builds a dependency graph. Upstream always runs first — nobody schedules anything by hand.

Tested

Grain, nulls, row counts: declared assertions run with the models. Failures surface in the pipeline, with the affected lineage visible.

Versioned

Every change is a reviewed pull request with history. “Who changed this and why” is always one click away.

It is still SQL against Snowflake. What changes is the starting point: instead of rebuilding demographics, registers and lookups for every piece of work, you join models that already exist, are already tested, and refresh on their defined schedules. The setup is a one-off; the everyday loop — edit, build, PR — takes minutes.

The handbook · reference for later

Longer-form reference pages on every topic the courses introduce — for when you need the full story on layers, sources, materialisations or snapshots, months after onboarding.

ConceptsWhy dbt, the data, layers, naming, design, production — 12 pagesWorking practiceSetup, sources, models, building, PRs, changing models, git recovery — 9 pagesGoing furtherThe dbt extension, macros, materialisations, clustering, snapshots, semantic views, Python — 8 pagesCommand referenceEvery dbt and git command, searchable and copy-paste ready