Choose the right visual for programme dataChoisissez le bon visuel pour les données de programme
A 20-minute guide to picking the chart that answers your question — with a quick-reference table and programme-specific rules you can use the next time you brief your programme.Un guide de 20 minutes pour choisir le graphique qui répond à votre question — avec un tableau récapitulatif et des règles spécifiques au programme à utiliser lors de votre prochain briefing.
Choose the right chart
The right chart depends on the question, not on the data you happen to have. Answer two or three questions below and the chooser recommends a chart, shows an example, and writes the prompt to build it from your downloaded programme file.
Make it read well
Avoid
Build it
- Download the IU-level dataset for your disease and country from your programme data portal (choose IU level, the disease, and all years; export as CSV).
- Set your country and disease below, then add the file to your Claude chat along with the prompt.
- Check the columns Claude plans to use before it builds, then compare the chart with the checklist above.
Quick reference
| If your question is… | Use | Programme example | Not |
|---|---|---|---|
| How has it changed over time? | Line chart | LF coverage 2014 to latest year | Bars for 10+ years; a second y-axis |
| Which is biggest or smallest? | Ranked bar | Districts by oncho coverage | Pie chart; unsorted bars |
| Did each one hit the target? | Bar with target line | Districts vs the 65% threshold | Red/green colour alone |
| How big is the gap per item? | Dumbbell | Population requiring vs treated | Two separate charts side by side |
| Where is it? | District map | Endemicity by IU, same-year boundaries | Rainbow colours; mismatched boundary years |
| How is the total split? | 100% stacked bar | Districts by endemicity status, per disease | Pie with more than 3 slices |
| When will we get there? | Trend with projection | Districts needing MDA, forecast to target | Forecast drawn like real data |
| What's the one number? | Headline number | People treated in the latest complete year | A one-bar chart |
| What was submitted when? | Status grid | JAP components × year | Colour without a symbol or label |
| Patterns across many districts and years? | Heatmap | IU coverage, 2016 to latest year | A line per district |
| Are two measures linked? | Scatter plot | Effective rounds vs coverage, per IU | Reading it as cause and effect |
Chart rules for programme data
- One y-axis per chart. Put coverage (%) and people treated (counts) on separate charts.
- Start percentage axes at 0 and draw the programme threshold as a reference line.
- Dash or flag the latest year if it looks incompletely reported.
- Use shades of one colour for ordered categories like endemicity; a different colour only when categories are unrelated.
- Label lines directly and always give a legend; never rely on colour alone.
- Put the table and years under every chart.
Not sure? Ask Claude to choose
Describe the question and audience and let Claude propose the chart before it builds anything.
I have attached the IU-level programme dataset for [COUNTRY] and [YOUR DISEASE]. I want to show [YOUR QUESTION] to [AUDIENCE]. Using only this file, recommend the best chart type and explain why, name one alternative you rejected and why, and list the columns you'd use. Wait for my go-ahead before building it.