Think & decide

Audit a chart for misleading design

Review scale, encoding, labels, missing context, and the claim against source values

7 min setupTested with:ChatGPTReviewed: 2026-08-28

Add your context

Your text stays in this browser. AILesson Prompts does not send it to a model or server.

Chart descriptionSource values and definitionsAudience and intended claim

Your prompt

Unfilled fields remain visible as placeholders, so you can still copy and edit the prompt

Audit this chart for misleading or unclear design.

Chart type, axes, scales, labels, colors, annotations, and title:
[chart]

Source values, definitions, units, denominators, periods, and missing data:
[source]

Audience and intended claim:
[claim]

Compare every visual encoding and textual claim with the source. Check truncated or unequal axes, area/volume distortion, dual axes, category order, omitted baselines, cherry-picked periods, changing denominators, missing uncertainty or sample size, color accessibility, and title wording. Rank findings as blocking, material, or minor. Then specify a corrected chart and state the strongest claim the evidence actually supports. Do not invent unseen source details.

Private by defaultPrompt assembly happens locally in your browser. Avoid placing confidential information into any AI service unless your organization allows it.

From input to outcome

A worked example

See how concrete context turns this recipe into a usable result

Actual input

Chart description: A 3D column chart titled 'New training doubles productivity'. It shows Before=82 and After=91 tasks per week. The y-axis begins at 80. Before is gray and After bright green. No sample size or dates are shown.
Source values and definitions: Same 12 employees measured for four weeks before and four weeks after training. Values are group means. No control group. Two employees were on leave during one after week. Individual variation and uncertainty are not supplied.
Audience and intended claim: Management slide intended to justify rolling the training out company-wide.

Example output

Blocking findings

The title is false: 91 is 11.0% above 82, not double. The y-axis starting at 80 makes a nine-task increase occupy almost the full chart height, and 3D perspective adds further visual distortion. Without a control group, the before-after difference cannot be attributed to training.

Material findings

The chart omits n=12, the four-week windows, paired measurement, leave-related coverage difference, and uncertainty. A mean alone hides whether improvement was widespread or driven by a few employees. Bright green also implies approval rather than neutral comparison.

Corrected specification

Use a paired dot plot of each employee's before and after mean if individual data is available, plus a clearly labeled group summary. Otherwise use a simple two-point slope or column chart with a zero baseline, flat 2D design, exact labels 82 and 91, and subtitle “Observed mean increased by 9 tasks/week (11%); n=12; uncontrolled before-after comparison.” Add the date windows, coverage note, and uncertainty once calculated. The strongest supported claim is that the observed mean was higher after training; the data does not establish that training caused the increase or predict company-wide results.

Why this works

1
Checking the visual against source values catches distortion that a style-only review misses.
2
Severity ranking separates truth-changing defects from ordinary polish.