Coming next: The Second Factory →

Back to Library
Data Foundations⚠️ Pitfall

The Goldilocks Data Model (Too Hot, Too Cold, Never Just Right)

The Problem

Two extremes plague analytics teams: (1) Data left unmodeled or haphazardly modeled, hindering scalability and usability, or (2) Data over-engineered, slowing progress and confounding even seasoned analysts. Fresh talent joins eager to build, only to hear "our data's a hot mess—expect to spend 70% of your time wrestling with it." Both extremes leave data harder to use, not easier. You can't achieve Analytics Powerhouse status without adhering to basic data modeling principles.

The Principle

A data model has ONE job: make data easier to use for those crafting solutions with it—the BI analysts and data scientists. Ask ruthlessly: "Does this model make the data easier to use—yes or no?" If yes, everything else (integration, quality monitoring, predictability, security) becomes way easier to pull off. Strive for "stupidly simple" data—so intuitive it's like picking a book off a shelf and instantly grasping its contents. Slow down just enough to speed up.

🎯

3 Steps to Take Action

  1. 1

    Ask the Brutally Simple Question: For every table/column, ask: "Does this make data easier to use?" If no, simplify it. If yes, document why.

  2. 2

    Onboard Test: Have a new analyst try to find data without help. Time them. If it takes >15 minutes to find common data, your model is too complex. Simplify.

  3. 3

    Build a Data Catalog: Document your tables, columns, relationships in plain English. Make it searchable. New people should find what they need in seconds, not hours.

In Practice

A Fortune 100 retailer had 847 tables in their data warehouse with no naming conventions, no documentation, and cryptic abbreviations. New analysts spent 6 months learning the maze before being productive. They rebuilt using simple principles: clear naming (sales_transactions, not sls_txn_f47), logical grouping by domain, one source of truth per metric. Onboarding time dropped from 6 months to 3 weeks. Data science team velocity tripled. The "stupidly simple" redesign took 4 months but paid back in saved time within 90 days.

More on Data Foundations

From the book

In the book, a new hire takes six months to become useful, mostly because of table names nobody can read. And there are five definitions of an active seller.

This card is one of 57 drawn from The Auction Block by Brian Cariveau — the story of an analytics leader with 90 days to fix what the cards describe.

Each leadership pattern falls into a few pitfalls most often. Find your pattern.The Analytics Leadership Mirror: free, 5 minutes, no account