Verified against ChatGPT · 2026-08-14
Turn a pile of open-ended survey comments into themes leadership can actually act on this quarter
Synthesizes raw open-text survey responses into a small number of concrete, prioritized themes with representative quotes and a recommended action per theme, instead of a vague summary that restates the survey question back at leadership.
The prompt
Ready to copy — highlighted parts are example details you can swap.
Analyze the open-ended survey responses below and produce a synthesis leadership can act on, not a restated summary of what the survey asked. SURVEY CONTEXT AND QUESTION ASKED Quarterly engagement survey, open text question: 'What's the biggest obstacle to doing your best work right now?' RAW OPEN-TEXT RESPONSES 'Too many meetings, I have 6 hours of meetings some days and no focus time.' / 'Unclear priorities since the reorg, don't know who owns what anymore.' / 'Meetings eat my whole calendar, hard to get deep work done.' / ... WHAT LEADERSHIP NEEDS TO DECIDE FROM THIS Whether to introduce a company-wide no-meeting day, restructure team ownership post-reorg, or both. KNOWN CONTEXT THAT MIGHT EXPLAIN A THEME The company completed a reorg two months ago that merged two previously separate product teams. Read through the raw responses and group them into no more than 5 themes, ranked by how many distinct respondents raised something in that theme, not by how emotionally strong any single comment sounds — a theme mentioned by 2 people should not outrank one mentioned by 12 just because the 2 people wrote more dramatically. For each theme, include 2-3 representative quotes pulled directly from the raw responses (verbatim, not paraphrased, so leadership can see the actual language people used) and a rough count or proportion of respondents who raised it. For each theme, propose one specific, concrete action leadership could plausibly take this quarter — not a vague direction like "improve communication," but something specific enough to actually be approved or rejected as a real decision. If known context is given that plausibly explains why a theme is showing up (a recent reorg, a tool migration, a leadership change), name that connection explicitly rather than presenting the theme as if it emerged from nowhere, since a theme presented without its likely cause reads as a mystery leadership has to re-diagnose from scratch. Note anywhere the responses genuinely conflict with each other (some people want more structure, others want less) rather than averaging them into a mushy middle recommendation that satisfies neither group. WHAT NOT TO DO Do not pad the synthesis with a restated methodology section explaining how you grouped things — lead directly with the themes. Do not editorialize beyond what the actual responses support; if a theme is thin (2-3 mentions), say so plainly rather than inflating it to sound more significant than the data shows. OUTPUT FORMAT Ranked themes (1 through up to 5), each with: proportion of respondents, 2-3 verbatim quotes, likely cause if known, and one specific proposed action. End with a short "Where responses conflict" note if applicable.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
The specific failure mode of a naive open-text synthesis is ranking themes by the vividness or emotional intensity of individual comments rather than by how many distinct people actually raised the underlying issue, which systematically overweights a small number of articulate or frustrated respondents and produces a synthesis that reflects who wrote the most memorable comment rather than what the broader population actually experiences — ranking explicitly by respondent count rather than comment intensity corrects for this. Requiring verbatim quotes rather than paraphrases matters because paraphrasing open-text feedback for a leadership audience tends to launder specific, often pointed language into vaguer corporate-safe phrasing, which is exactly the softening that causes leadership to underestimate how strongly a theme is actually felt — the real words people used are the evidence, and paraphrasing loses exactly the signal that should inform how seriously to take it. Forcing every theme's proposed action to be specific enough to be approved or rejected as a real decision, rather than a vague direction like "improve communication," targets the common way survey syntheses fail to produce change: a report that ends in generalities gives leadership nothing concrete to say yes or no to, so it tends to be acknowledged and then shelved, whereas "introduce a company-wide no-meeting day" is a specific proposal someone can actually approve this quarter. Surfacing genuine conflicts between respondents rather than averaging them into a mushy middle addresses a real analytical trap: if half the respondents want more structure post-reorg and half want less, a synthesis that proposes "moderately adjust structure" satisfies neither group and obscures that there are actually two distinct needs in the population that may require two different responses, not one compromise position.
What you get back
Theme 1 (raised by ~40% of respondents): Meeting overload crowding out focus time. Quotes: 'I have 6 hours of meetings some days and no focus time.' Likely unrelated to the reorg — this is a pre-existing calendar-culture issue. Proposed action: pilot a company-wide no-meeting Wednesday for one quarter. Theme 2 (~25%): Unclear ownership since the reorg. Quotes: 'Don't know who owns what anymore.' Likely cause: directly tied to the team merger two months ago. Proposed action: publish an updated RACI for the merged teams within 2 weeks.
Verified against
ChatGPT GPT-5.1 · 2026-08-14
Changelog
- 2026-08-14 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
