Summarize engagement survey comments into themes
Free-text survey comments are grouped into themes, counted by team and quoted so leaders can see what people said in their own words. Small groups are suppressed to protect anonymity. For HR teams reading thousands of comments before the leadership readout.
Function: HR · Industry: Any industry
Summarize engagement survey comments into themes
6 steps
At a glance
- Trigger
- A survey closes and results are exported.
- Typical inputs
- Survey responses with team and tenure fields
- Anonymity thresholds
- Previous survey themes
- Question wording
- Systems it usually connects to
- Google Sheets
- Notion
- Slack
- Snowflake
- How the work splits
2 rules, 2 model, 2 person steps
How it runs, step by step
6 steps, each tagged by what does the work. Rules are fixed logic, a model reads or drafts, and a person decides.
- Rules Fixed logic in code
- Model AI reads or drafts, with a confidence score
- Person A named person decides
- 1
Prepare the data
RulesResponses are cleaned, and groups smaller than your anonymity threshold are merged or hidden before any analysis.
- 2
Group comments into themes
ModelA model proposes themes from the comments and assigns each comment to one or more of them.
- 3
Review the themes
PersonAn HR analyst edits theme names, merges overlaps and checks a sample of comments in each.
- 4
Count and compare
RulesTheme frequency is calculated by team and compared with the last survey, using the same definitions.
- 5
Pull representative quotes
ModelA model selects short quotes that illustrate each theme, with identifying details removed.
- 6
Prepare the readout
PersonThe HR lead decides what goes to leaders and what actions to propose.
Where people approve
The decisions that stay with your team, whatever the confidence score says.
- The HR lead approves quotes before they are shared.
- Anonymity thresholds are set by HR and cannot be lowered by the tool.
- Leaders decide on actions.
KPIs to track
What to measure, before and after. Targets come from your own baseline.
- 01Days from survey close to readout
- 02Share of comments assigned to a theme
- 03Analyst hours per survey
- 04Actions agreed per team and completed
Questions about this workflow
Can the model identify who wrote a comment?
It should not be asked to. We remove names and obvious identifiers before analysis and suppress small groups. Distinctive writing can still hint at an author, which is why quotes are reviewed by a person before they leave HR.
How do you make sure themes are accurate?
An analyst reviews each theme and a sample of its comments. Theme definitions are kept between surveys so changes over time reflect the comments, not a relabeling.
Which steps use rules, a model and a person?
Rules: prepare the data; count and compare. Model: group comments into themes; pull representative quotes. Person: review the themes; prepare the readout. Rules are fixed logic in code. A model reads or drafts and reports its confidence. A named person decides.
What do we need to start?
Typically: survey responses with team and tenure fields; anonymity thresholds; previous survey themes; question wording. You also need read access to Google Sheets, Notion, Slack and Snowflake, or the equivalents you use. A workflow audit measures your volumes and past cases first, then gives a fixed price for the build, so you can decide with real numbers.