Research

Analyze survey ratings and comments in Excel

Combine rating distributions with traceable comment themes and response-bias limits

8 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.

Survey question and decisionFields, scales, and sampleNumbered anonymized comments

Your prompt

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

Design and report an Excel analysis of survey ratings and open comments.

Survey question and decision:
[question]

Fields, rating scale, invitations, responses, and groups:
[structure]

Numbered anonymized comments:
[comments]

Keep rating analysis and comment coding linked but distinct. Report counts, distribution, median or mean only when appropriate, missingness, response rate, and group sample sizes. Create a small theme codebook, allow multiple themes per comment, attach comment IDs as evidence, and distinguish frequency from importance. Do not generalize respondents to all invitees without discussing nonresponse. Return supported findings, contradictions, minority views, and claims the data cannot support.

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

Survey question and decision: After a new onboarding session, how useful did participants find it, and what should be changed before the next cohort?
Fields, scales, and sample: 120 invitees; 48 responses. Rating 1=not useful to 5=very useful; optional Comment; Cohort is NewHire or InternalTransfer. Ratings: 1:2, 2:5, 3:12, 4:20, 5:9. NewHire n=34, InternalTransfer n=14. Comments completed by 30 respondents.
Numbered anonymized comments: C01: Clear examples, but setup section was too fast. C02: Good overview; I needed more time to practice. C03: Most content repeated what internal transfers already know. C04: The checklist was useful. C05: Audio failed for ten minutes. C06: I still do not know who approves access. C07: Practice exercise made the process click. C08: Too long overall, especially the company introduction.

Example output

Response rate is 48/120 = 40%, so results describe respondents, not all invitees. The rating distribution is 4.2% at 1, 10.4% at 2, 25.0% at 3, 41.7% at 4, and 18.8% at 5. The median is 4 and 60.4% selected 4–5; report the full distribution rather than only a mean. Compare cohorts only with their separate distributions and n because InternalTransfer has 14 responses.

Initial codebook: Pace/practice (C01, C02, C07), relevance by cohort (C03), useful job aid (C04), delivery failure (C05), unclear ownership/process (C06), excessive length (C08). C07 is positive evidence for practice while C02 requests more of it; these are complementary, not contradictory. C03 is a minority but decision-relevant cohort concern.

Supported actions: slow the setup section, add practice time, clarify access approval, fix audio readiness, and review whether the introduction can be shortened or separated for transfers. Do not rank these solely by eight comments; code all 30 and show theme count plus rating distribution of commenters. The data cannot establish why 72 invitees did not respond, whether the session improved job performance, or whether changes caused future rating differences without a comparison design.

Why this works

1
Separating numeric distribution from coded comments prevents anecdotes from replacing the rating evidence.
2
Comment IDs and a codebook make qualitative themes traceable and reviewable.