The six sessions 📊 · one dataset, all the way through
The running dataset is Daybreak's subscription numbers - signups, cancellations, revenue by roast, retention by cohort. The coffee-subscription company from the SQL, system-design, product-design, communication and course-design courses. Same numbers every session, so the only thing that changes is the decision you made about them.
What is the question?
The chart's job decides its form: magnitude, identity, polarity, change, or one headline number. Plus the answer people find hardest to accept, which is that quite often it should not be a chart.
📐 Session 2 · easyChoosing the form
Bar, line, dot, heatmap - picked by the job rather than by habit. Sorting, small multiples, pie charts honestly assessed, and the one rule with no exceptions: never a dual axis.
⚖️ Session 3 · mediumHonesty
The lie-factor bench. Truncated axes, area against length, 3D, gradients and cherry-picked windows - with the distortion computed from the rectangles the chart actually drew.
🎨 Session 4 · mediumColour that works
The four jobs colour can do, one rule each. Colour-vision deficiency through a real matrix transform, contrast as arithmetic, and why four hues for four bars of the same measure is a mistake.
🔎 Session 5 · hands-onMarks and anatomy
Thin marks, recessive scaffolding, direct labels instead of a legend, text in text colours, the hover layer, and the table view that is not a consolation prize.
📤 Session 6 · hands-onPublishing it
The same chart in a report, a slide, a phone and print. Dark mode as a second palette rather than an inversion, forced colours, and what a screen reader actually gets.
Your passport 🎫
Clear all three quiz questions in a session and it gets stamped. Six stamps means every chart you make from now on has a job and can be checked. Stamps live in this browser only.
What this course is not, so you land in the right place 🧭
| If your question is | Go here instead |
|---|---|
| How do I read a metric and question it? | learn-data-literacy-with-phoebe |
| How do I commission and govern dashboards? | learn-business-intelligence-with-phoebe |
| What does the statistic actually mean? | learn-statistics-with-phoebe |
| How do I turn a set of charts into a story? | learn-visual-storytelling-with-phoebe |
| Which encoding does this question need, and is the picture telling the truth? | you are in the right place |
Session 6 is deliberately not about dashboards. That is business-intelligence's job, and two courses teaching one subject disagree with each other eventually.
What is honest here 🧾
Unusually for a course in this estate, almost everything in the signature widget is measured rather than modelled.
| Claim | Status |
|---|---|
| The lie factor | Really measured off the drawn geometry - truncate the axis and the number moves because the rectangles moved. Blunt by design: it compares the largest and smallest item, so a chart can score 1.00 and still mislead by its window or its denominator, which session 3 covers separately |
| Contrast | Really measured. WCAG 2.1, computed on the chart's own text against its own surface |
| Colour-vision separation | A real matrix transform and an OKLab distance - but a simulation of dichromacy, not what a person with that condition sees. Good enough to catch a palette that will fail |
| The chart palette | The validated categorical default. Lightness, chroma, CVD separation and normal-vision floor all pass; the contrast warning on two slots is discharged by direct labels, which the chart carries |
| Daybreak's numbers | A teaching dataset. The measurement code is real |
The knowledge map 🧠
The whole course at a glance - hover a session to spotlight its concepts, click any node to jump in.