The Segmentation tab lets you analyze all of your employee responses to the sentiment questions asked in your courses, sliced and diced by various learner attributes—such as Department, Tenure, or Job Group. This view highlights how different groups experience workplace topics differently, revealing both strengths and potential risk areas across your organization.
Please see the video below for a walkthrough of the tool:
What does the Segmentation by Question report show me?
Each row in the table represents a segment (such as an individual leader, department, or tenure group). Each column represents a question from the selected catalog topic.
The percentage values indicate how many learners in that segment selected a healthy response—a response that aligns with a positive, respectful, or safe workplace experience. For negatively worded questions (for example, "I have observed bullying on my team"), the percentage shown represents those who disagreed with that statement.
-
Average % Healthy: An overall indicator summarizing how positively a segment responded across all questions in the view. Averages are weighted by response count across questions within a question set—so questions with more respondents have proportionally more influence on the average than questions with fewer respondents.
Segments with insufficient question coverage are excluded from averaging. If a segment answered fewer questions than the median across all segments for the selected catalog and time period, the Average column will display "Insufficient coverage" rather than a score. This prevents a partial sample from being mistaken for a full assessment. -
Color Shading: Highlights segments that differ from the overall organizational sentiment.
Red or pink indicates below-average sentiment, while green indicates above-average sentiment.- In general, darker greens (and generally higher numbers) = more positive sentiment than the overall organization. Darker reds (and generally lower numbers) = less positive sentiment than the overall organization
- Note: Colors are based on how much a segment's responses deviate from the organizational average, not on the raw score itself. (See "What do these colors mean?" below for a technical explanation). As a result, you might see:
- A 45% healthy score shaded green if the organization overall scored lower, or
- An 88% healthy score shaded red if the organization overall scored higher.
- Sorting: Hover over a question header to reveal sorting controls. Click to sort by highest or lowest scoring segment for that question.
- Trend Arrows (▲ / ▼): A small arrow next to an Average % Healthy score indicates how that segment's score changed compared to the same period one year prior—for example, Calendar Year 2025 vs. Calendar Year 2024, or the Nov 2024–Apr 2025 period vs. Nov 2023–Apr 2024. ▲ means the score improved; ▼ means it declined. No arrow means there was no meaningful change, or no prior-year data exists yet for that segment. Arrows are not shown on rows displaying "Insufficient coverage."
- Blank cells: A blank cell in a question column means fewer than 5 learners in that segment answered that specific question—below the minimum threshold required to display a score. This is different from "Insufficient coverage," which appears only in the Average column and reflects whether the segment answered enough questions overall to produce a reliable average.
How do I use the segmentation controls?
Use the filters at the top of the dashboard to refine which data appears in your segmentation table. The controls are divided into three main steps:
Step 1 – Choose Time Period and Content
-
Response Range:
- Calendar Year: A fixed range from January 1 to December 31 for the selected year.
- Rolling Six-Month Period: Six-month periods starting on the first of the month and ending six months later. This time period was chosen to reflect the cadence of most training campaigns, which typically span less than six months.
- Date Range: Displays the specific dates covered by your selection (for example, Jan 01 2025 – Today).
- Question Set: Choose the course or topic you want to analyze (for example, Preventing Workplace Harassment or Workplace Violence). You can also select All Deployed Titles to see questions across all courses in a single view.
These filters determine which learner responses are included in your dataset.
Step 2 – Choose Population
- Segment Results By: Choose a learner attribute to break down responses by group—for example, Department, Tenure, or Senior Leader.
-
Minimum # of Learners: The minimum learner threshold helps you narrow down the results by number of learners who have responded to questions. This filter enables you eliminate smaller groups and focus on larger populations, as desired.
- We require a minimum of 5 respondents as the threshold for showing scores. This is to protect learner confidentiality, so that learners feel comfortable responding candidly.
- We recommend setting your minimum threshold to .5% of your learner responses to eliminate noisy smaller segments. This recommended threshold is shown in the tooltip to the right of the filter.
-
Include New Hires: Toggle on to include learners classified as New Hires (less than one year of tenure) in score calculations, or off to exclude them. This is off by default because new hire responses often differ significantly from the established workforce, and including them can mask trends in your broader employee population.
How "New Hire" is determined: If your organization provides hire dates, tenure is calculated directly from those. If hire dates are unavailable, Emtrain infers tenure from lesson activity: a learner is classified as a New Hire if their first recorded lesson falls within the past 12 months and after your organization's platform was fully established. Learners whose tenure cannot be confidently determined are shown as Unreported.
Note: If Emtrain has not received hire date or tenure data for your organization, this toggle will have no effect.
These options define which population segments appear in the table and help ensure your analysis is statistically meaningful.
Step 3 – Add Additional Filters
-
Risk Area: Filter questions to a specific HR or compliance risk category:
- Harassment & Discrimination
- Hiring, Performance Management, & Terminations
- Wage & Hour
- Workplace Safety
- HR Reporting & Retaliation
- Code of Conduct
- Data Privacy & Information Security
- Bribery & Corruption
- Antitrust & Fair Competition
- Reporting & Whistleblowing
-
Culture Skill: Filter questions to a specific culture competency:
- Advancing Allyship
- Being Well
- Cultivating Authenticity
- Deciding Together
- Demonstrating Integrity
- Embracing Interculturalism
- Encouraging Empathy
- Engaging Morally
- Ensuring Equality
- Fostering Curiosity
- Instilling Accountability
- Managing Power
- Mitigating Bias
- Nurturing Trust
- Thinking Systemically
- Valuing Diversity
-
Show Unreported: Include the Unreported row in the heatmap. "Unreported" represents learners whose value for the selected attribute (e.g., Department) is missing or was not provided to Emtrain. It only appears for dimensions where at least one learner does have a known value—dimensions with no data at all are omitted entirely.
Note: A segment value may also show as Unreported if the group contains fewer than 5 learners, to protect individual confidentiality. -
Question Keyword / Segment Keyword: Use these search boxes to filter for specific questions or segments. For example,
- Search Segment Keyword for "marketing"
- Search Question Keyword for "respectful behavior"
These filters let you refine your analysis to focus on specific risk areas, culture competencies, or key themes.
The summary panel in the top-right corner of the dashboard provides quick context for your current selections:
- Current Period Selected: The exact date range being analyzed based on your Step 1 selections.
- Total Learners In This Period: The number of unique learners who answered at least one sentiment question during the selected period.
- Learner Attribute Data Last Refreshed: When your organization's HR attribute data (department, tenure, etc.) was last updated. Responses in the current period will reflect any changes made since this date; past closed periods are unaffected.
Why do you hide some segments by default?
By default, segments that answered fewer questions than the median for the displayed question set are automatically hidden to avoid cluttering the screen and to prevent misleading comparisons. For example, if most segments answered 8 out of 10 questions but one segment only answered 3, that segment is hidden because its Average % Healthy would reflect a very different—and smaller—slice of the question set than everyone else.
The number of hidden segments is displayed inline above the heatmap (e.g., "1 segment(s) hidden"). Check the box next to that indicator to reveal all segments, including those with lower question coverage.
What do the colors mean?
Colors represent segments that are trending above or below the overall organization sentiment across all segments. This shading helps highlight where specific groups' experience differ from the broader organization.
For a more technical understanding, the darkest colors (red and green) indicate values that are greater than +/- 2 standard deviations from the overall organizational sentiment for that specific segment, question, and time period combination.
Lighter shades represent smaller differences, starting at +/- 0.5 standard deviations and stepping incrementally toward the darkest tones as deviation increases.
The color shown in the Average % Healthy column summarizes the overall sentiment from all questions in the current view, using the average of the colors derived from individual questions.
Interpreting results (example)
In the example screenshot for Acme Corporation, the results are segmented by Senior Leader for the Preventing Workplace Harassment course. Each row represents responses from an executive's team, and columns show key questions like "I have observed bullying on my team" or "My co-workers are respectful to others regardless of their race, gender, or identity."
- Leaders like Casey Lang (CHRO) and Quinn Lopez (SVP, Marketing) show higher overall healthy responses on average (78% and 76%), reflecting stronger team culture.
- Segments such as Taylor Reid (CFO) and Avery Morgan (COO) appear in red or pink for several items, suggesting lower confidence or higher perceived conflict within their teams.
These insights help identify where additional communication, leadership development, or culture interventions may be valuable.
Best practices
- Use consistent time periods when comparing data across quarters or years to ensure valid trend comparisons.
- Review both low and high scoring groups to understand areas needing attention as well as culture strengths.
- Focus on the Average % Healthy column first to identify overall trends ("this department responded less positively on average") and then hone in on specific questions based on those.
- Tip: If the Average % Healthy column only shows gray, then this means that every segment trended similarly to the overall organization sentiment.
How are learner attributes matched to the selected time period?
Learner attribute data (department, location, tenure, etc.) is tracked historically. Each time a learner's attribute changes, the previous value is preserved along with the dates it was active. When you select a time period in Step 1, each learner's responses are matched to whichever version of their attributes was current during that period.
This means a learner who changed departments mid-year will appear under their previous department for responses recorded before the change, and under their new department for responses recorded after.
How uploads affect historical periods: Uploading new learner attribute data does not retroactively change past completed periods. For example, a file uploaded on May 1st will not alter how learners were attributed during the November 1st – April 30th rolling period—that period is locked and will always reflect the attributes that were on file while it was active. However, any in-progress period (one that includes today) will update to reflect the most recent upload, since it has not yet closed.
The Learner Attribute Data Last Refreshed date shown in the top-right corner of the heatmap indicates when this historical record was last updated from your HR system or file import.
Frequently asked questions
Does selecting a Question Set filter by the version of the course a learner completed?
No. The Question Set filter controls which questions are displayed in the heatmap—it does not filter responses by course version. All responses from all versions of a course are included regardless of which version a learner completed.
What does "Insufficient coverage" mean in the Average column?
It means a segment did not answer enough questions to produce a reliable average. Averages are weighted by response count across questions, and if a segment answered fewer questions than the median across all segments for that catalog and time period, the Average column will show "Insufficient coverage" rather than a score. This prevents a partial sample from being mistaken for a full assessment.
What does "Unreported" mean?
Unreported appears when a learner's value for the selected attribute (e.g., Department) is missing or was not provided to Emtrain. It only appears for dimensions where at least one learner does have a known value—dimensions with no data at all are omitted entirely. A segment value may also show as Unreported if the group contains fewer than 5 learners, to protect individual confidentiality. Use the Show Unreported toggle to include or exclude this group from the heatmap.
What does "New Hire" mean and how is it determined?
New Hires are learners with less than one year of tenure. If your organization provides hire dates, tenure is calculated directly from those. If hire dates are unavailable, Emtrain infers tenure from lesson activity: a learner is classified as a New Hire if their first recorded lesson falls within the past 12 months and after your organization's platform was fully established. Learners whose tenure cannot be confidently determined are shown as Unreported. Use the Include New Hires toggle to include or exclude this group from score calculations.
Why do some questions show trend arrows (▲ / ▼) but others don't?
Trend arrows compare a segment's current score to the same period one year prior. If a question did not exist in last year's version of a course—because it was added, reworded, or removed in a subsequent version—there is no prior-year score to compare against, so no arrow is shown. This is expected behavior, not missing data. As questions persist across multiple annual cycles, arrows will begin to appear for them.
Where does this learner attribute data come from?
You! Please work with your Admin and your Emtrain team to send relevant learner attribute data. Learn more about how to work with Emtrain on importing segments.