As of 13 August 2026, AI can choose the right chart for your data.

What to actually do

1. Hand it to a person

   The route this page recommends

   A person who owns the outcome does this end to end, worth it when the failure is dear.

2. Use a tool built for this

   Second choice

3. Do it yourself

   The distant third

   A chat interface, chat-fluent skill, and roughly 15 minutes until you can act on the result.

   How to actually do it

   1. Open the spreadsheet or CSV and note the column headings, units, date range, row grain and any missing or duplicated records.
   2. Write down the single question the chart must answer, such as comparing categories, showing change over time, showing a distribution or examining a relationship.
   3. Paste the relevant data or a representative extract into the prompt, then add the column definitions, intended audience and decision the chart will support.
   4. Ask the model to recommend one primary chart, alternatives, axes, grouping, labels, filters and any data preparation needed.
   5. Create the recommended chart in Excel, Google Sheets or your reporting tool using the stated fields and aggregation, then compare its values against the source data.
   6. Check the axis scale, totals, date ordering, category labels, missing values and outliers, and ask a colleague to confirm that the chart answers the original question without implying more than the data shows.

   Prompt

   Choose the most appropriate chart for the data below.
   
   Data:
   [PASTE THE DATA, OR A TABLE WITH COLUMN HEADINGS]
   
   Column definitions and units:
   [EXPLAIN WHAT EACH COLUMN MEANS]
   
   Question the chart must answer:
   [STATE THE BUSINESS QUESTION]
   
   Audience:
   [STATE WHO WILL READ IT]
   
   Decision or action the chart will support:
   [STATE THE DECISION, OR WRITE "DESCRIPTIVE REPORTING ONLY"]
   
   Requirements:
   - First identify the variable types, such as categorical, numerical, date or geographic.
   - Recommend one primary chart and explain exactly why it answers the question.
   - Give up to two credible alternatives and explain when each would be better.
   - State what should be on the x-axis, y-axis, colour, labels and filters.
   - Check for missing values, duplicated rows, unsuitable aggregation, extreme values and misleading scales before recommending the chart.
   - Do not invent values, trends, relationships or column meanings. Ask concise questions where essential information is missing.
   - Say when a chart is not appropriate and a table or another analysis would be safer.
   - Keep the recommendation suitable for the stated audience and decision.

   Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.