The challenge
The HR team had been running a quarterly pulse in a generic survey tool. Two problems kept surfacing. First, leadership skimmed the executive summary and skipped the data because it was not clear what the numbers meant. Second, several department heads quietly stopped trusting the tool after a small-team rollup felt identifiable to a known dissatisfied employee.
Both problems pointed at the same gap. The instrument and the report did not carry their own credibility. The HR team needed an engagement tool with anonymity protections built in and a report that explained itself to non-researchers.
How they designed the survey
The team adopted ReliCheck's Employee Engagement template and turned on three protections by default. K-anonymity suppression hides any rollup with fewer than 5 respondents, showing only a count. Anonymous mode strips respondent identifiers from the dashboard. Manager-level access controls limit each manager to their own direct rollup plus the all-staff number.
The instrument is 22 items across five drivers (purpose, autonomy, growth, recognition, and team). Three open-ended items ask what is working, what should change, and one optional manager-feedback item that managers themselves never see.
What the data showed
First run: 1,094 of 1,400 employees responded (78%). The engagement composite came in at α = 0.91 with KMO 0.87. Two of the five drivers (recognition and growth) scored more than half a point below the others on a 5-point scale, with consistency across teams.
The board-ready report opened with the composite, the alpha tile, and one paragraph of AI plain-language summary. Driver-level breakdowns followed, with team rollups suppressed below the k-threshold so no team felt exposed. AI theme extraction on the open-ended items grouped 312 free-text responses into 6 themes with example quotes attached to each.
"Our CFO read every page. That had not happened with the prior tool. Reliability statistics in the header gave the report an authority that the prior dashboards never had, and the AI summary saved me an entire afternoon of write-up."
At a methods glance
| Sample size | n = 1,094 across 9 departments and 27 teams |
| Instrument | 22 Likert items, 5 drivers, 3 open-ended |
| Reliability | Composite α 0.91; driver αs 0.83 – 0.89 |
| Anonymity | k = 5 minimum on every rollup; no rollup below threshold ever shown |
| AI themes | 6 themes from 312 open-ended responses, manually reviewed |
| Export | PDF for the board, Excel for HR business partners, manager-specific PDFs |
What they did with the result
- The executive team narrowed a list of 12 candidate action items down to 4, choosing actions tied to the two lowest-scoring drivers (recognition and growth).
- Each manager received a personalized PDF report covering their own team rollup, plus the all-network composite for comparison. Managers below the k-threshold received the all-network report only.
- The HR team scheduled the next pulse 90 days out so the same instrument could measure movement on the two flagged drivers. Reliability statistics carried over, so the team had a defensible before-and-after comparison instead of just two snapshots.
Read more about how hr and teams teams use ReliCheck →