The six sessions 🎯 · from "can we predict it" to a brief she can read
The running case is Harbourline, an invented chain of 38 pharmacy and clinic outlets across Singapore and Malaysia, and its operations director Adeline Tan, who has never heard the word pipeline. Every fact on her sheet is printed in the bench's bank file; the course claims the mechanism, never the numbers as facts about pharmacies.
The gap, from the scientist's chair
Why "can we predict it" is the most dangerous request a scientist gets, the outcome door opened widest, and the three facts no model can be scoped without: the exact target, how often it happens, and what each wrong guess costs.
🗂️ Session 2 · coreWhat has been recorded, and for how long
Door 4 for a scientist: history, labels and what was never written down. At Harbourline nobody records a stock-out and the stock sheet is overwritten every Friday, so the first project is recording, not modelling.
🎛️ Session 3 · bench nightThe scientist's questions, and the bench
Five kinds of question and the same operations director run five ways. The fast interview scores 15 and writes down "forecast sales"; the doors in order 46; outcome first, 55. Then you take the chair.
💬 Session 4 · coreTeaching while you ask
The one concept worth ninety seconds in a scientist's meeting: the two wrong guesses cost different amounts. Twelve teach-back lines rewritten for a person who has never seen a confusion matrix.
🌱 Session 5 · coreThree projects, and the one that is not a model
Her pains become three candidates, and the first is a daily stock-out log in four outlets. Why the non-ML fix comes first, and what a model needs before it can beat a pharmacist looking at a shelf.
📄 Session 6 · capstoneThe mock meeting and the one-page brief
A full run on the bench with your own sheet, scored, then the brief she can read: the target in her words, the baseline, what you will not build yet. Final scorecard.
Your passport 🎫
Clear all three quiz questions in a session and it gets stamped. Six stamps means you can walk into a meeting with a business person who knows nothing about data, leave with the decision, the current way and the real data state, and send a brief she can read the next morning. Stamps live in this browser only.
Four chairs at the same table 🪑
The same framework, the same company and the same bench engine, built four times for four roles. Each course marks a different door to open widest and carries its own question bank and fact sheet. The framework itself lives in the data analyst course's session 1; this course links to it.
What this is not, so you land in the right place 🧭
| If your question is | Go here instead |
|---|---|
| I am the executive. How do I read a number with confidence and brief an analyst? | Data Literacy ↗ |
| I have been handed a question. How do I sharpen it and say what the data can and cannot answer? | Data Thinking ↗ |
| How do I present findings, handle a room, and influence a decision? | Communication ↗ |
| How do I run product discovery as a discipline, week after week? | Product Thinking ↗ |
| How do I frame a decision properly, with options, evidence and a call? | Decision Intelligence ↗ |
| What is a data maturity model, and how do I run a real assessment? | Governance 101 ↗ |
| I am the data person. I have 45 minutes with a business person who knows nothing about data, and I need to leave with something I can build. | you are in the right place |
The first meeting, run by the data side. The decision, the current way and the real data state, got out of a person in words that teach her something as she answers. Reading numbers, sharpening a handed question, presenting, and discovery as a product discipline are each taught once, in the course that owns them, and linked from here rather than repeated.
Choose your path 🗺️
Three doors into the same conviction: the way of asking, not the person across the table, decides what you leave with.
What is honest here 🧾
The company is invented and says so on every page that uses it. The bench is counted from a printed file. Where a page rests on a book or a paper, it says which and how it was read, and where a widely quoted number could not be traced, the page says that instead of quoting it.
| Claim | Status |
|---|---|
| Harbourline, Adeline Tan, and every fact on her sheet | Constructed. Written to be typical of a mid-size retail chain; printed in full in the bench's bank file; never a claim about a real business |
| Every count on the bench: net value, false beliefs, doors, the 20-minute counter | Really computed in your browser from the printed bank, and verified headlessly in node before any page quoted a number |
| Tappers and listeners: about 50 percent predicted, 3 of 120 named | Reported from Heath and Heath, Made to Stick (2007), citing Newton's 1990 Stanford dissertation, which was not read at source |
| Measurement as a reduction of uncertainty about a decision | Hubbard, How to Measure Anything (2014), definition as it appears across summaries; the book was not opened here |
| Ask about the past, not the future; people say yes to be polite | Fitzpatrick, The Mom Test (2013). Only his verified wording is quoted; a paraphrase that circulates as a quote is not |
| The analytics translator role and its predicted headcount | Role named from the 2018 HBR article; the headcount figure was not read at source and is not quoted |
The knowledge map 🧠
The whole course at a glance - hover a session to spotlight its concepts, click any node to jump in.