Employee listening
Turn employee comments into themes you can examine.
Coding gives each relevant passage a descriptive label so a team can compare accounts systematically. Use this worksheet to retain the example, its context and the reason for the label. The worked example shows how several accounts can lead to a focused check without turning comments into a workforce verdict.
Set the analysis question and the unit
Start with the work question the comments were collected to inform. Define a unit such as one passage describing a specific event. Keep its participant reference, collection question and permitted source location. One response may contain several events, and several comments may describe the same event. Record those relationships before counting. Work within the access arrangements explained to participants.
Read the whole account before applying labels
A sentence can change meaning when the surrounding answer explains the situation. Read the response and relevant question, then extract the passage that relates to the analysis question. Note whether the person describes their own experience, something they heard or a suggestion. Preserve the original wording in its approved location. The worksheet can carry a short permitted paraphrase where broader access to the original would be inappropriate.
Build a small codebook
Begin with a few labels that describe the work, such as access unavailable, approval owner unclear or instruction changed. Write a definition, an inclusion rule and an exclusion example for each. Leave room for a new label when an account does not fit. A broad label such as communication can hide several different issues, so test whether the passages actually concern the same step.
Calibrate with another reviewer
Ask another authorized reviewer to code a small common set independently. Compare the passages selected and the reasons for each label. A difference may reveal an unclear definition, two issues combined in one sentence or a missing piece of context. Agree the revision and apply it consistently to earlier passages. Keep a dated codebook version so the analysis can be checked later.
Worked example from a fictional onboarding inquiry
Six fictional colleagues describe how they found an approver during their first week. One could not open the directory. Another opened it but found two possible names. A third knew the right owner and received a prompt answer. A fourth used an old checklist, while another had no task that needed approval. The last colleague was unsure which directory the interviewer meant. The rows below preserve these differences rather than reducing every answer to poor communication.
Keep the contrary account beside the pattern
The colleague who reached an approver promptly helps define the conditions under which the process worked. Ask what source, task and work period their example covered. Their account may fit the same conditions as the difficult cases or expose a meaningful difference. Keep it visible in the theme review. A code should organize the account without forcing it into the explanation the sponsor expected.
Count against an explicit opportunity to answer
Count unique participants separately from passages when describing who raised an issue. Record how many participants were asked about the topic and whether they had experienced the relevant task. In this fictional set, one access problem is one reported access problem. The colleague who did not encounter an approval task is not evidence that access worked. None of these six accounts estimates the experience of the wider workforce.
Write the finding and the open check separately
A defensible summary of the example is that participants described several different barriers to locating an approver, alongside one successful case and two accounts needing different treatment. The next check could inspect the current directory, checklist and new-starter access route. “Managers are unhelpful” would go beyond the passages. Give a process owner the specific records to inspect and the unresolved questions they need to answer.
If AI helps, retain the review trail
Use an approved tool only with material it may receive. Keep proposed codes distinguishable from accepted codes, and inspect the supporting passage for each material finding. Check missed contrary accounts and passages assigned to a broad label. Record which codebook version and human decisions shaped the output. A fluent theme description still needs the same contextual review as a manually written summary.
Use the worksheet as an analysis aid
The Framework Method describes a systematic approach to comparing cases and themes, developed in qualitative health research. GOV.UK’s analysis guidance also distinguishes observed material, interpretation and action. This original workplace worksheet adapts those organizing ideas for a bounded employee-feedback question. For a complex or consequential inquiry, involve someone with appropriate qualitative research experience and preserve the limits of the collection method in the final brief.
Frequently asked questions
Questions that come up
How many codes should we use?
Use enough to distinguish the work issues relevant to the question. Review broad or overlapping codes and retain a route for material that does not fit. The examples on this page are a starting point for one fictional inquiry.
Can we turn code counts into an engagement score?
Code counts describe the collected material under a stated coding rule. Keep the question, opportunity to answer, participant base and coverage visible. A new score would require its own definition and evidence for the use you intend.
What if two reviewers disagree?
Return to the passage and its context. Compare the code definitions, record the reason for the difference and revise the codebook where needed. Keep a genuinely unresolved interpretation visible in the review note.
Talk to Lontra HR
Bring one real HR question.
See how a first campaign could work for your team.