Segment survey results

What this achieves

Segmenting shows the same survey broken down by one dimension at a time, so you can see which parts of the workforce differ from the company average. It is where a survey result stops being a number and starts telling you where to look.

Required role: Administrator.

Steps

Go to Engagement → Surveys.
Open the survey.
Open the Segments tab.
Choose a Dimension.
Read the table, then change the dimension and read it again.

The dimensions

Option	Description
Department	Results per department.
Grade	Results per grade.
Tenure	Results by how long people have been with the company.
Gender	Results by gender.

Look at more than one. A difference that appears by department may really be a difference by tenure, if one department has done most of the recent hiring.

Example: Engineering at HC Corp UK Ltd scoring lower than Operations tells you something. Both departments’ newest joiners scoring lower than everyone else tells you something different, and the second reading is the one that changes what you do.

What the table shows

Option	Description
Segment	The department, grade, tenure band, or gender.
Responses	How many responses that segment contributed.
Engagement	The engagement score for that segment.
eNPS	The eNPS for that segment, where the survey has an eNPS question.

Read the response count on every row before you read the score beside it. A segment with a handful of responses produces a number that moves several points if one person answers differently, and it will look like a real change next quarter when it is not.

When the product declines to segment

Where there are too few responses to break down, the view says there are not enough responses to segment rather than showing you a table.

That is a floor, not your policy. It stops the most obvious disclosure; it does not decide what is safe to publish at your sizes, and the threshold is not shown on screen. Set your own minimum, above whatever the product enforces, and apply it before anyone sees the table.

Example: a department of six at HC Corp UK Ltd may well produce a segment row. Whether you should put that row in front of anyone is a different question, and the answer does not change because the product was willing to draw it.

If a dimension you need refuses to segment, the honest options are to widen the population on the next wave or to accept that this question cannot be answered at that grain. Re-cutting the same data a different way until something appears is not one of them.

Gender and other sensitive dimensions

Gender is available as a dimension and is among the most identifying at small sizes, because the smaller group in a segment can be very small indeed even where the segment as a whole is not.

Decide whether you will report on it at all, decide it once, and decide it before you have seen a result you want to publish.

Cross-checking against the questions

The segment view gives you the score per segment, not which questions moved it. When a segment stands out, go back to the Results tab: the question distributions are what tell you whether the issue is about managers, recognition, workload, or pay.

Example: Operations at HC Corp UK Ltd scoring low overall is a prompt. Operations scoring low specifically on the recognition questions is something you can act on.

What happens next

A segment worth acting on belongs with whoever owns that part of the business, alongside the response count and the question detail that explains it. The engagement dashboard shows the same department breakdown across surveys rather than for one, which is where you check whether a segment is having a bad quarter or a bad year.