Prompt Library

ChatGPT Prompts for Charts and Graphs That Communicate

20 copy-paste prompts

20 copy-paste ChatGPT prompts for charts: type selection, design principles, common mistakes to avoid, color, annotation, and the workflows that turn data into clear visual communication.

Chart Type Selection

4 prompts

Type from Question

1/20

For [question + data], best chart type. Output: recommendation + alternatives + when each. Common: bar (compare), line (trend), scatter (correlation), pie (parts, ≤4 slices), table (specifics). Match visual to question.

Selects chart types.

💡

Pro tip: Wrong chart type = obscures data. Bar > pie for compare (almost always). Pie acceptable only for ≤4 slices showing parts of 100%. Match question, not data.

Bar vs Column

2/20

Bar chart vs column chart. Output: bar (horizontal, good for long category names, ranked lists) vs column (vertical, good for time series, fewer categories). Different uses.

Bar vs column.

💡

Pro tip: Bar (horizontal): long labels readable, good for ranked top-N lists. Column (vertical): time series, fewer items. Default to column unless labels long; then bar.

Line vs Area

3/20

Line chart vs area chart. Output: line for: trend, multiple series, precise values. Area for: cumulative totals, parts of whole over time, single series. Different cognitive loads.

Line vs area.

💡

Pro tip: Line = precise trend. Area = mass + flow. Multiple lines OK; multiple stacked areas = often confusing. Default to line; area for cumulative narratives.

Scatter vs Bubble

4/20

Scatter vs bubble plot. Output: scatter for 2-variable correlation. Bubble adds 3rd dimension via size. When 3-variable matters, bubble; when 2 enough, scatter.

Scatter vs bubble.

💡

Pro tip: 2-variable correlation = scatter. Adding size dimension (e.g., revenue + growth + market size) = bubble. 3rd dimension via size; humans struggle with 4+.

Prompts get you started. Tutorials level you up.

A growing library of 300+ hands-on AI tutorials. New tutorials added every week.

Start 7-Day Free Trial

Design Principles

4 prompts

Chart Junk Removal

5/20

[Describe chart]. Remove chart junk: 3D effects (avoid), gridlines (minimal), legends (only when needed), labels (minimum useful), colors (purposeful only). Less = more.

Removes chart junk.

💡

Pro tip: Default chart software adds: gridlines, dark borders, legends, 3D effects, fancy colors. Strip back to data + minimal context = professional. Tufte's data-ink principle.

Title-as-Takeaway

6/20

Chart title-as-takeaway. Output: not "Q3 Revenue" (topic) but "Q3 revenue grew 30%" (takeaway). Title carries the message.

Writes takeaway titles.

💡

Pro tip: Topic titles = "what is this chart about?" Takeaway titles = "here's the point." Titles should pass standalone test: read just titles, get the story.

Axis Honesty

7/20

Honest axes. Output: y-axis usually starts at 0 (truncating misleads), exception (data range tight), label units, axis breaks if needed (clearly marked). Truncation = manipulation accusation.

Sets honest axes.

💡

Pro tip: Truncated y-axis can mislead (small change looks huge). Default: start at 0. Exception: data range very tight (e.g., 99.5%-99.7% uptime). Then break clearly marked.

Color Choice

8/20

Color in charts. Output: 1 highlight color + neutrals (gray) > rainbow palette. Color encodes meaning; not decoration. Categorical (distinct colors) vs sequential (gradient) vs diverging (red-blue).

Chooses chart colors.

💡

Pro tip: Rainbow chart = "everything important" = nothing important. Highlight + gray = "this matters." Color encoding intentional. Most charts over-color.

Common Mistakes

4 prompts

Pie Chart Misuse

9/20

Pie chart misuse. Output: >5 slices unreadable, comparing 2 pies hard (use stacked bar), 3D pies distort, similar-sized slices indistinguishable. Most pie charts should be bars.

Avoids pie chart misuse.

💡

Pro tip: Pie charts work only for: ≤4 slices, parts of 100%, single comparison. Almost always: bar chart better. Default to bar; reach for pie only when explicitly justified.

Dual-Axis Charts

10/20

Dual-axis charts (two y-axes). Output: when justified (related metrics in different scales), when misleading (manipulation easy), alternatives (small multiples, indexed values).

Avoids dual-axis pitfalls.

💡

Pro tip: Dual-axis charts can mislead by manipulating either axis. Reader can't tell what's "high" or "low." Often better: small multiples (separate charts) or normalize to indexed values.

Time-Series Issues

11/20

Time-series chart issues. Output: irregular intervals (mark them), missing data (show gaps, not interpolation), date format consistency, x-axis density. Time-series special.

Avoids time-series issues.

💡

Pro tip: Time-series with missing data interpolated = misleading. Show gaps explicitly. Irregular intervals (different spacing) = stretches reality. Treat time honestly.

Comparison Trap

12/20

Comparison trap: 2 different time periods on same chart with different scales. Output: comparison must be apples-to-apples (same units, same scale, same time periods). Otherwise misleading.

Avoids comparison traps.

💡

Pro tip: Comparing absolute numbers across periods of different lengths = wrong. Normalize: per-day, per-customer, indexed to base. Apples-to-apples mandatory.

Like these prompts? There are full tutorials behind them.

Learn the workflows, not just the prompts. 300+ easy-to-follow tutorials inside AI Academy — and growing every week.

Try AI Academy Free

Specialty Charts

4 prompts

Small Multiples

13/20

Small multiples (grid of small charts). Output: when useful (comparing many series, breakdowns), how to design (consistent scales, sorted logically), advantages over single chart with many series.

Builds small multiples.

💡

Pro tip: Small multiples = many small charts side-by-side. Beats single chart with 20 lines (spaghetti). Each small chart focuses; brain compares grid. Underused; powerful.

Heatmap Design

14/20

Heatmap for [data]. Output: when useful (matrix of values, calendar of frequency), color scale (sequential), labeling, alternative if heatmap noisy. 2D categorical visualization.

Designs heatmaps.

💡

Pro tip: Heatmap = matrix shaded by value. Useful for: weekly patterns by hour, correlation matrices. Color scale matters: sequential for magnitude, diverging for above/below mean.

Sankey Diagram

15/20

Sankey diagram for [flow data]. Output: when useful (flow visualization, conversion funnel, energy flow), design considerations, tools (D3, online tools, BI built-in). Specialized.

Designs Sankeys.

💡

Pro tip: Sankey = flow visualization. User journey through funnel, energy flow, budget allocation. Powerful when flow matters. Don't use for non-flow data.

Bullet Chart

16/20

Bullet chart (KPI vs target). Output: design (single bar with target marker + range bands), use cases, why better than gauge for KPIs.

Designs bullet charts.

💡

Pro tip: Bullet chart > gauge chart for KPIs. Gauges hard to read precisely. Bullet shows: actual, target, performance bands. Few BI tools have built-in; worth learning.

Frequently Asked Questions

No directly. AI describes chart structure + suggests type. You build in tool (Excel, Sheets, BI tool). For visualization gen: use tools with native AI (Gemini, Copilot, BI tool AI features).

Bar chart. Most universal. Compare anything. Master bar (including stacked, grouped, ranked) = handles 60% of visualization needs.

Brain bad at comparing angle/area. Bar charts use length (best). Pies acceptable for ≤4 slices showing parts of 100%; otherwise bar wins. Pie = often default but rarely best.

"Storytelling with Data" (Cole Knaflic) for business context. "Visual Display of Quantitative Information" (Tufte) for principles. "Functional Art" (Cairo) for journalism. All change how you see charts.

8% of men, 0.5% of women color-blind. Red/green encoding fails for them. Combine color with: shape, label, position. Or use color-blind-safe palettes (ColorBrewer, Viridis). Accessibility = good design.

Prompts are the starting line. Tutorials are the finish.

A growing library of 300+ hands-on tutorials on ChatGPT, Claude, Midjourney, and 50+ AI tools. New tutorials added every week.

Start 7-Day Free Trial Explore AI for Data Analysts

7-day free trial. Cancel anytime.

Browse All Prompt Collections