As of 13 August 2026, AI can analyse your staff survey results.

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 30 minutes until you can act on the result.

   How to actually do it

   1. Export the survey responses to a spreadsheet and remove names, email addresses, employee numbers, exact timestamps, free-text details and other combinations that could identify a person.
   2. Open a separate document and paste the survey purpose, questions, response options, relevant group definitions, comparison period and any known limitations.
   3. Check the anonymised spreadsheet for blank cells, duplicate responses, inconsistent group labels and accidental identifying details, then correct or remove them.
   4. Paste the context and anonymised data into a chatbot using the supplied prompt, or upload the spreadsheet to Julius AI and give it the same instructions.
   5. Compare every reported response count, percentage, average and chart with the spreadsheet, checking the denominator and missing-value treatment for each result.
   6. Ask a colleague who understands the survey and workplace context to challenge the themes, group comparisons, quotations and proposed actions against the original questions and anonymised responses.
   7. Rewrite the report so that evidence, interpretation, limitations and proposed follow-up actions are clearly separated, then share only the anonymised version with the intended audience.

   Prompt

   Analyse the anonymised staff survey data below. Do not identify or guess the identity of any respondent, and do not infer sensitive personal characteristics. Use the survey questions and response scale exactly as provided. First report the number of responses and missing values for each question. Then calculate the relevant counts and percentages, showing the denominator for each percentage. For rating questions, report the distribution, average only where it is meaningful, and any important limitation. For free-text responses, group comments into clearly named themes, state the number of comments in each theme, include only short anonymised excerpts, and separate direct evidence from your interpretation. Compare groups only where each group has enough responses to make the comparison useful, and flag small or uneven groups instead of drawing conclusions. Identify the strongest positive and negative patterns, changes from any comparison period supplied, contradictions and results that need further investigation. Do not claim causation. End with a concise report containing findings, evidence, limitations, practical follow-up questions and actions that do not assume the analysis is correct. Use plain British English and invent nothing. If the data or context is insufficient for a conclusion, say so.
   
   Survey purpose and decisions this will inform: [paste here]
   Survey questions and response options: [paste here]
   Comparison period or benchmark, if available: [paste here]
   Relevant group labels, with no identifying details: [paste here]
   Anonymised survey data: [paste here]

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