ChatGPT Prompts for Survey Analysis: 12 for Real Results

ChatGPT prompts for survey analysis work best when you paste your real counts and comments and ask for tables, comparisons, and themes that show their base sizes. The 12 prompts below follow the order of a real analysis: check the data, build tables, compare groups, code open answers, and write the readout. Always check the math in the output against your survey tool before you share it.

Privacy note: Remove names, emails, phone numbers, and other personal details before you paste any results or comments, and follow your survey's consent terms and your company's data rules.

Check and clean the data

1. Totals check

Here are my survey counts by question and option [paste]. Add up each question, compare the totals, and list anything that does not reconcile or looks like a skip pattern.

2. Cleaning plan

My export has these columns [list] and these problems [duplicates, blanks, test responses]. Write a step-by-step cleaning plan for [Excel or Google Sheets], with the formula or filter for each step.

3. Question type map

Here is my questionnaire [paste]. Label each question as single choice, multi-select, scale, ranking, or open-ended, and say how each type should be summarized.

Build results tables

4. Results table per question

Turn these counts [paste] into a table per question with count and percentage to one decimal. Put the base n above each table and combine top-two and bottom-two boxes for scale questions.

5. Multi-select handling

Question [text] allowed several answers. Here are the counts [paste] and the number of respondents [n]. Show percentages of respondents, explain why they add to more than 100%, and rank the options.

6. Segment comparison

Here are results for [question] split by [segment] [paste counts]. Show percentages within each segment with each base, and flag any segment with fewer than [number] people as directional only.

Code open-ended answers

7. Draft codebook

Read these open-ended answers [paste]. Propose 4 to 8 themes, each with a one-line rule for what counts, and one example comment per theme.

8. Apply the codebook

Using this codebook [paste], tag each comment below [paste] with one or more themes. Return a table of comment number and themes, then the count per theme.

9. Quote picker

For each theme [list], pick the two clearest verbatim quotes from these comments [paste]. Do not edit the wording, and skip any quote with personal details.

Write the readout

10. Key findings

From these tables and themes [paste], write 5 key findings for [audience]. Tie each finding to a number with its base, and keep each under 30 words.

11. Caveats section

My survey went to [who] by [method] and got [n] responses. List the main limits of these results in plain language, including who might be missing.

12. Next survey questions

The decision we need to make is [decision], and these results left these gaps [list]. Suggest 4 follow-up questions, with answer options, that would close them.

Run the whole analysis in one prompt

The flagship Survey Analysis Prompt: Results Tables and Open-End Themes takes your survey goal, questions, counts, segment splits, open comments, audience, and a minimum base. It returns a data check, results tables, segment comparisons with flags, an open-end codebook with quotes, key findings, and caveats. Its example analyzes a community pool survey about opening hours.

Tips for accurate survey analysis with AI

Paste counts, not just percentages, so every number can be checked. Remove names and emails from comments first. Ask for a significance test only when you have raw counts per group, and verify any test in a spreadsheet or stats tool before you report it.

Related prompts

For deeper coding of comments, use the survey open-end codebook prompt. Before your next survey, run the survey questionnaire with bias checks, and for interviews and usability studies see ChatGPT prompts for UX research.

Find more in the Research category, look through Browse, or post your own on Share.