Description
The problem was never the SQL. It is that everyone writes their own. Marketing has one definition of an active user, finance has another, and the CEO’s dashboard has a third, so three people walk into a meeting with three revenue numbers and spend the meeting arguing about whose is right instead of deciding anything. Every new dashboard is someone re-deriving the same metric slightly differently, and the differences hide until they surface in front of leadership. You have become the human source of truth: nothing ships until you personally check it, which means you are the bottleneck and the report factory at once, copying the same filter into query after query, terrified of the day two of them silently disagree. You know how to write the query. What nobody taught you is how to define a metric one time, in one place, so that every dashboard, every explore, and every export computes it the same way, and how to hand people the freedom to answer their own questions without handing them the freedom to invent their own numbers. That is the gap between writing reports and owning the numbers a company runs on.
This book takes you from the person who writes every report to the person who owns the one definition everyone else builds on. It starts where the pain starts, in the meeting where two dashboards disagree, and builds the idea the whole book turns on: a metric should be defined once, in a modeled layer, and computed the same way everywhere it appears. You will build one real model across the book, for an online fitness-app company called Cadence, and take it from a pile of conflicting queries to a governed layer the whole company self-serves from. You will define dimensions and measures once in LookML so ‘active user’ and ‘monthly revenue’ mean exactly one thing, wire up the joins so an explore lets a non-technical colleague ask their own question without writing SQL, and build dashboards and reports people run themselves instead of queuing behind you. You will see why a governed metric beats everyone’s own number, and how to keep the layer honest as definitions change. Then you will make it reach further with Looker Studio, the free tool that turns a modeled source or a raw table into a shareable dashboard in minutes, with the sharing, embedding, and permission controls that decide who sees what. By the end, a metric means one thing, people answer their own questions, and when someone asks why revenue is what it is, the answer is in the model and not in your head. For the analyst who wants the whole company to trust one set of numbers, and wants to be the reason it can.
Who this book is for
This book is for: the analyst or analytics engineer who has become the report factory, who fields the same ‘why does your number differ from mine’ argument every week, who is copying the same CASE WHEN into a fifth dashboard, and who wants a way for the whole company to self-serve reliable numbers without every one of those numbers running through their inbox.
What makes this one different
The 5 Steps to Becoming the Person Who Owns the Numbers the Whole Company Decides On. The method that turns the person who writes every report into the person who owns the single definition everyone else builds on. When the whole company self-serves its own numbers, the hard part stops being SQL and becomes agreement: a metric has to mean one thing everywhere, people have to answer their own questions without inventing their own totals, and the numbers have to hold as definitions change. You build one real modeled layer from conflicting queries to a governed source of truth, and learn how to define a metric once so every dashboard and export computes it the same way, how to let colleagues explore without writing SQL, and how to reach everyone fast and free with Looker Studio.
- Define a metric once and reuse it everywhere
- Let colleagues ask their own questions without SQL
- Give people dashboards they run for themselves
- End the everyone-has-their-own-number arguments
- Reach everyone fast and free with Looker Studio
What’s in this book
- Chapter 1: The meeting where two dashboards disagreed and I had no answer
- Chapter 2: I keep copying the same metric into every report. How do I define it once?
- Chapter 3: People keep asking me for numbers. How do I let them ask the data themselves?
- Chapter 4: How do I get out of the report factory without losing control of the numbers?
- Chapter 5: Everyone has their own definition of revenue. Whose is right?
- Chapter 6: I need a dashboard in front of the whole company by tomorrow, for free
- Chapter 7: The day nobody asked me whether the number was right
