Rebuilding a chart every time someone questions it can cost you an afternoon. Most charts fail for the same handful of reasons, and spotting them takes minutes once you know what to look for. These five Claude prompts give you a repeatable critique, from the axes up to the title, so you fix the real problem the first time.
Why charts get challenged
- The y-axis starts above zero, which exaggerates small differences and invites a fair challenge.
- Labels are missing or cryptic, so the reader has to guess what a color or a bar actually means.
- The chart type does not match the message: a pie chart for ten categories, or a line chart for data with no order.
The prompts
1. Full critique checklist
Critique this chart: [chart description]. The message I want it to land is [intended message], and the audience is [audience]. Go through it in order: title, axes, labels, color, and chart type. For each, say whether it helps or hurts the message, and give one fix.What it does: Runs a full pass so nothing obvious gets missed.
How to use: replace [chart description], [intended message], and [audience]; keep the rest as is.
2. Axis and scale review
Review the axes on this chart: [chart type] showing [data range]. Tell me whether the scale hides or exaggerates the difference I am comparing, and whether the axis starts at zero. Suggest the setting that keeps the comparison honest.What it does: Fixes the most common way charts mislead without meaning to.
How to use: replace [chart type] and [data range]; keep the rest as is.
3. Color and label review
Look at the colors and labels in this chart: [chart description]. The audience is [audience] and it will appear in [medium]. Flag any color that carries meaning I did not intend, any label a reader could misread, and any legend that could be removed by labeling directly.What it does: Cleans up the details that slow readers down.
How to use: replace [chart description], [audience], and [medium]; keep the rest as is.
4. Chart type fit check
I have this data: [data description]. The message I want to show is [message]. Is the chart I chose the right one, or should I switch? Compare it with two alternatives, and tell me which one makes [message] easiest to see at a glance.What it does: Confirms the chart type fits before you spend time polishing it.
How to use: replace [data description] and [message]; keep the rest as is.
5. Rewrite the title and labels
Here is my current chart title: [current title]. Rewrite it so it states the takeaway, not the topic, for [audience]. Then suggest short labels for the elements in [data] that match. Keep it plain and free of jargon.What it does: Turns a descriptive title into one that states the point.
How to use: replace [current title], [audience], and [data]; keep the rest as is.
Better questions to ask
| Weak ask | Better ask |
|---|---|
| Is this chart good? | Critique this bar chart for [audience]. Check whether the y-axis starting above zero exaggerates the gap, and suggest a fix that keeps the comparison honest. |
| Fix my chart title. | Rewrite [current title] so it states the takeaway for [audience], not the topic. Keep it under 12 words. |
Both rewrites point the model at one specific problem. A chart critique is only useful when it tells you what to change, not just that something feels off.
Keep going
- More Data & Analysis prompts
- ChatGPT prompts for data analysis plans
- Gemini prompts for ROI calculations
- Claude prompts for SERP analysis
More free prompts for Data & Analysis are waiting in the GuPrompt library.