AI Prompts for Engagement Survey Analysis

Survey results are usually reported as averages and heatmaps, which tell leaders where scores are low and nothing about what to do. The useful analysis asks which themes are most associated with the outcomes that matter (intent to stay, overall experience), where the differences between groups are real and where they are noise, what the comments say when read carefully, and which two or three actions would matter most.

These prompts do the driver analysis, read the comments, and build the action plan. They absorb the library's engagement-initiatives prompt into the action step, where it belongs — after the analysis, not instead of it. Anonymity rules apply throughout: no reporting below the threshold, no attempts to identify individuals from comments.

Before you use these

Have these ready to replace the highlighted [variables]:

• Results by item and theme, overall and by reportable group
• Outcome items (intent to stay, recommend, overall)
• Open-text comments (with any identifying details already removed)
• Response rates by group and the anonymity threshold

The prompts

• 1. Run driver analysis and honest segmentation
• 2. Read open-text comments into themes with evidence
• 3. Turn results into a small set of actions with owners

1. Run driver analysis and honest segmentation

Act as a people analytics partner analyzing engagement survey results for [company / population].

Results: [theme and item scores overall and by reportable group; outcome items; response rates and group sizes; prior period if available]
Anonymity threshold: [minimum n]

1. Driver analysis: which themes are most associated with the outcome items (correlation or a comparison of outcome scores between high and low scorers on each theme); rank them; state the strength and the limitation (association, not cause; self-report; same-time measurement).
2. Priority matrix: themes by importance (association with outcomes) × current score; the quadrant that matters (important, low score).
3. Group differences: for each reportable cut, whether the difference is large enough given sample size and response rate to act on; classify as real / probable / noise; never report below the threshold.
4. Change from the prior period: where scores moved and whether the movement is meaningful.
5. Response-rate bias: groups with low response rates and how that limits interpretation.
6. What the data cannot answer and the follow-up (focus groups, comment analysis) that could.

Present as a short narrative with tables. Do not present a heatmap of averages as analysis.

2. Read open-text comments into themes with evidence

You are analyzing open-text survey comments for [company / population].

Comments: [paste de-identified comments]
Quantitative themes: [list]

1. Code each comment to one or two themes (from the survey's themes, plus new ones that emerge); count by theme.
2. Sentiment and specificity per theme: share negative/neutral/positive; share that name a specific, actionable issue versus a general feeling.
3. Representative quotes per theme (2–3): choose ones that convey the point and cannot identify the author — remove or generalize any detail that could (team names, projects, roles held by few people, incidents); flag any comment that should be routed to HR under policy (safety, harassment, misconduct) rather than reported as a theme.
4. New themes: issues not asked about in the survey, with counts — candidates for the next survey or immediate attention.
5. Comparison with the scores: themes where comments contradict or sharpen the quantitative picture.
6. The five most actionable specifics in the comments.

Never attempt to identify who wrote a comment. Report themes, not individuals.

3. Turn results into a small set of actions with owners

Act as a people-ops lead at a [size] company with [context, e.g., hybrid teams, low scores on recognition and clarity] turning survey analysis into action.

Driver analysis and comment themes: [summary: the important, low-scoring themes and what comments say]
Constraints: [budget, capacity, what leaders have committed]

1. For each priority driver, propose actions sorted into zero-budget, low-budget, and invest. For each: what it is, the mechanism by which it moves the driver, effort, expected effect, and how to measure it. Prioritize manager behavior and system fixes over events and perks; avoid pizza-party clichés.
2. Select two or three organization-level actions with owners and dates — no more.
3. Team-level: the guidance for managers to run a results conversation with their team and choose one team action; the template.
4. Measures: the survey items and the operational indicators that will show whether each action worked, and when.
5. Communication of results and actions: what is shared with whom, at what level, by when; the honest statement of what will not be addressed and why.
6. Follow-up pulse: the two or three items to re-ask in [n] months.

Fewer actions, done and visible, beat a long plan that proves the survey was ignored.

A priority matrix, illustrated

Themes plotted by association with intent-to-stay and by current score, for a 400-person company. Illustrative.

Related prompts

• Employee Survey Design
• Employee Retention Analysis
• Exit Interview Analysis
• Employee Communications and Change Announcements
• Manager Coaching
• HR Analytics and People Metrics

Logical next step

After this, most HR teams move on to Employee Survey Design.

All Human Resources prompts · Search the full library