A key driver is a survey question that has a strong statistical relationship with the overall outcome, or focal point, being measured. In an employee engagement survey, the focal point is usually the engagement index.

Key driver analysis helps organisations understand which aspects of employee experience are most closely associated with engagement. It provides an evidence-based starting point for deciding where to investigate and where action may be most valuable.

A key driver is not necessarily one of the highest- or lowest-scoring questions. A question may score poorly but have a relatively weak relationship with engagement. Another may have a moderate score but a much stronger relationship. Looking at both the score and the driver strength provides a fuller picture.

Key drivers are determined through statistical analysis of survey responses. The platform identifies which eligible survey questions have the strongest relationship with the focal point score.

Key drivers may vary between departments, teams, locations and demographic groups. For example, development opportunities may be strongly associated with engagement in one group, while recognition or leadership may have a stronger relationship in another.

These differences reflect patterns within each group’s responses. They do not necessarily mean that one factor is universally more important than another.

Key drivers can help organisations prioritise where to explore further. A low-scoring question with a strong relationship to engagement may warrant particular attention, while a high-scoring key driver may represent an important strength to maintain.

A useful way to interpret the results is:

A question does not need to be a key driver to be important. Organisations may reasonably take action on issues because they affect employee experience, reflect a strategic priority, or are particularly relevant to a local team.

Key drivers are based on the responses collected for a particular survey and reporting group. They may therefore change between survey rounds as people’s experiences and organisational circumstances change.

Key driver analysis identifies statistical relationships, not cause and effect.

A strong relationship between recognition and engagement does not prove that improving recognition will, by itself, cause engagement to increase. Both measures may be affected by other factors, and related survey questions may capture overlapping aspects of the employee experience.

The analysis also does not explain why a question is associated with engagement or prescribe a particular action. That requires organisational knowledge, professional judgement and discussion with employees.

Key drivers should therefore be considered alongside:

Key driver analysis is one lens on the data, not the whole story.

Should you have access to the Key Drivers' report, you will also be able to look at the correlation to any of a survey's themes, not just the focal point.

The focal point is the outcome against which other survey questions are assessed. For an employee engagement survey, this is usually the engagement index.

The engagement index is calculated as the average response across its constituent engagement questions. It is treated as the focal outcome measure for the analysis.

A focal point could also represent another defined outcome, depending on how the survey and dashboard have been configured.

The platform uses Pearson correlation analysis to measure the linear relationship between the focal point and each eligible survey question individually.

For each question, the system calculates Pearson’s correlation coefficient, r.

Questions with the strongest positive correlations are ranked as the strongest key drivers. The top five are displayed as key drivers in the dashboard, while the wider ranked results are available elsewhere in the reporting.

The coefficient is represented by the on-screen driver-strength indicator and is also provided numerically in the results download.

Mathematically, Pearson’s r ranges from -1 to +1:

The dashboard presents driver strength using a positive 0-to-1 scale because the key driver view ranks positive relationships with the focal point.

A higher positive coefficient indicates that respondents who score highly on the question also tend to score highly on the focal point, and respondents who score poorly on the question also tend to score poorly on the focal point.

It does not indicate how much the focal point would change if the question score changed.

The analysis is recalculated for reportable organisational and demographic groups where the configured anonymity threshold is met. This allows different patterns to be identified across teams, departments, locations and other groups.

However, the anonymity threshold is a confidentiality safeguard rather than a test of statistical reliability. Meeting the reporting threshold does not, by itself, establish that a correlation is precise, stable or statistically significant.

Results for smaller groups should therefore be interpreted with particular care.

The platform calculates a separate Pearson correlation between each eligible question and the focal point. It does not currently use multivariate regression within the dashboard analysis.

This means the analysis:

A high correlation should therefore be interpreted as evidence of a comparatively strong relationship within the available dataset, not as proof of influence or a guaranteed route to improving engagement.

Key drivers are most useful as evidence-based prompts for investigation and prioritisation. They indicate where further attention may be worthwhile, but the decision to act should reflect the complete evidence base.

This includes the strength of the relationship, the question’s current score, benchmark position, changes over time, qualitative feedback, local priorities and whether the issue is practically actionable.

Used in this way, key driver analysis helps organisations move beyond simply identifying their lowest scores. It supports better questions, more focused conversations and more informed action, while remaining clear about what the statistical analysis can and cannot establish.