Contenu du pack

Un fichier SKILL.md (le guide pratique) + un modèle de départ. Aperçu :

Paired Column Chart

When to use

• You need to compare two series across the same categories — this year vs. last year, plan vs. actual, group A vs. group B
• A before/after story where each category has a "before" number and an "after" number
• A slide or report where the audience must see which series wins in each category at a glance
• You have a table with one category column and exactly two numeric columns to compare
• A dashboard panel where two metrics share categories and deserve equal visual weight

Workflow

1. Shape the data (input: raw table → output: tidy comparison table)
   Reduce the input to exactly three columns: one category column and two numeric columns with plain-language headers ("Planned" and "Actual", not "Series1" and "Series2"). One row per category, categories in a meaningful order — chronological, ranked by the gap, or ranked by size — never alphabetical by default. If the raw data has more than two numeric columns, split it into multiple charts rather than cramming extra series into the pairs. If a category is missing one of the two values, decide explicitly: show a gap marker, or drop the category — never silently plot zero.
2. Set the scale (input: tidy table → output: axis spec)
   Start every value axis at zero — a truncated axis turns a 5% gap into a visual landslide and readers will not trust the chart again. Pick a round-number step (5, 10, 25, 50, 100) so gridlines land on numbers people can do arithmetic with. If the two series live on wildly different magnitudes (one in tens, one in thousands), do not force them onto one axis: normalize to an index, use per-unit values, or pick a different chart. State the unit once in the axis title ("Revenue, $ thousands"), not inside every label.
3. Pair the bars (input: axis spec → output: chart layout)
   Place the two bars for each category adjacent with a small gap between the pair and a wider gap between categories, so the eye groups each pair first and scans across pairs second. Use two clearly distinct fills — one neutral (grey/silver) and one accent — with a legend that sits above the chart, not buried below it. Keep bars the same width, keep the gap between paired bars constant, and never use 3D effects, gradients, or shadows that change a bar's apparent height. If there are more than about 10 categories, switch to horizontal bars so the category labels stay readable.
4. Label directly (input: chart layout → output: labeled chart)
   Put the value on or just above each bar whenever there are 10 or fewer pairs — readers should not have to trace a bar back to the axis to know its number. Format values consistently: same decimal places, same thousands separators, same unit handling across both series. Rotate category labels only as a last resort; prefer shorter labels, wrapping, or horizontal bars. Every number a reader needs must be on the chart itself, not hidden in a footnote.
5. Annotate the story (input: labeled chart → output: finished chart)
   Add at most two or three annotations, each pointing at the comparison that matters: the biggest gap, the one reversal where the usual winner loses, the category that drives the total. An annotation states the fact and the size ("Actual beat plan by 18% in March"), not a vague reaction ("Great month!"). Finish with a title that states the finding ("Actual beat plan in 7 of 10 months"), a subtitle with the scope ("Q1–Q3, all regions"), and a source line. Then check the whole chart at thumbnail size: if the comparison is not obvious at a glance, the chart is not done.

Aperçu du modèle

Paired Column Chart — Template

Fill in the three columns, then hand the table to the chart builder.

Data table

Category order: {chronological / ranked by gap / ranked by size} Missing values: {show gap marker / drop category} — never plot as zero

Axis spec

• Unit: {e.g. Revenue, $ thousands}
• Axis starts at: 0 (always)
• Gridline step: {5 / 10 / 25 / 50 / 100 — round number}
• Series magnitudes: {similar — one axis OK / very different — normalize or split}

Layout

• Orientation: {vertical (≤10 categories) / horizontal (11+ categories)}
• Series A fill: {neutral grey/silver}
• Series B fill: {accent}

Le pack de compétence lui-même (SKILL.md + modèle) est rédigé en anglais.

Acheter le pack complet — $4.99 Pro — les 511 compétences