Free guide for analytics engineers
Ready for the Gate
A free 22-page guide to the analytics engineering practices a semantic layer depends on. Seven chapters, each a set of short rules with a check you can tick and the usual way it goes wrong.
- Start from a contract, not a ticket: purpose, grain, owner and tests agreed before any SQL
- Declare the grain, then defend it: fan-out, duplicate keys and conformed dimensions
- Test what the business relies on, reconcile to the ledger, and keep what was known when
- Put definitions in the layer, and change them without breaking what they mean
- Where AI helps in the repo, and where it mustn't decide
- Every check on four printable checklist pages
Who it's for
The people who build what the semantic layer stands on
A semantic layer can only govern what reaches it. If the grain is wrong, a definition was settled in someone's SQL, or history was quietly restated, the layer serves the mistake to every dashboard and every AI assistant at once.
This guide is for analytics engineers and the heads of data who review their work. It isn't a tour of analytics engineering. It covers only the practices a governed definition relies on, as rules you can check.
Want the full argument?