What is key driver analysis in an employee engagement survey?
Key driver analysis is a way of ranking the drivers of employee engagement, such as communication, growth and recognition, by how closely each one moves with an outcome across your teams. Where teams that score higher on a driver also score higher on engagement, that driver is a lever. Where they do not, it is not, however low its score.
Key driver analysis: ranking engagement drivers by how strongly they move with engagement or eNPS across groups, so effort goes where it can change the outcome.
A score tells you where a driver stands. Its link to the outcome tells you whether improving it is likely to matter. Key driver analysis puts the two together, which is why it is the slide leaders act on in a results presentation.
Why is the lowest score often the wrong place to start?
The lowest score is the most visible number in a report, so it tends to become the priority. But a driver can score low across every company and still have little to do with how engaged people are. CultureMonkey's benchmark data shows how often the two lists disagree.
Most closely linked to eNPS
| Driver | Link (r) | Score rank |
|---|---|---|
| Communication | 0.54 | 16 of 22 |
| Growth & Development | 0.53 | 18 of 22 |
| Meaningful Work | 0.52 | 8 of 22 |
| Purpose Alignment | 0.51 | 2 of 22 |
| Involvement | 0.51 | 13 of 22 |
| Work Environment | 0.48 | 12 of 22 |
Lowest scores
| Driver | Score | Link (r) |
|---|---|---|
| Rewards | 7.14 | 0.38 |
| Recognition | 7.63 | 0.43 |
| Work-Life Balance | 7.66 | 0.35 |
Link is the Pearson correlation between a driver's score and eNPS across benchmark segments: 0.5 or more is strong, 0.35 to 0.5 moderate. See all 22 drivers
Communication and growth and development move eNPS more than any other driver, yet rank 16th and 18th of 22 on score. Rewards scores lowest of all, but its link to eNPS is only moderate. A plan built on the lowest scores would put rewards first and miss the two biggest levers.
How do you run a key driver analysis?
To run a key driver analysis, pick one outcome, score every driver and the outcome for each group, correlate each driver with the outcome across groups, check you have enough groups, plot score against impact and name the zones. The six steps below work in a spreadsheet; leadership reports in CultureMonkey do them automatically.
- Pick one outcome
Choose engagement or eNPS as the result you want to move, and keep it the same for the whole analysis.
- Score every driver for every group
Calculate each driver and the outcome for each team, location or manager group that clears your anonymity threshold.
- Correlate each driver with the outcome across groups
For each driver, measure how closely groups that score higher on it also score higher on the outcome. Leave the driver out of the outcome it is compared with, so it cannot inflate its own link.
- Check you have enough groups
With fewer than about eight groups, a correlation is too unstable to rank anything. Say the analysis cannot tell yet rather than print a ranking.
- Plot score against impact
Put the link to the outcome on one axis and the driver score on the other, and split both at the median.
- Name the zones and pick one or two
Label each driver Protect, Fix first, Watch or Maintain, then choose one or two Fix first drivers for the year, not the whole list.
Groups must clear your anonymity threshold before they enter the analysis. More groups make the ranking steadier, but never at the cost of reporting a group small enough to identify someone.
How do you read the four zones?
Split drivers at the median link and the median score and each one lands in one of four zones. Fix first and Protect are the leadership agenda; Watch and Maintain are not.
| Zone | Where it sits | What it means |
|---|---|---|
| Fix first | Strong link, score below the median | Moves engagement here, and currently scores below your median. The best candidates for the year's effort. |
| Protect | Strong link, score above the median | Moves engagement here, and is already strong. Losing it would cost you, so keep investing. |
| Watch | Weak link, score below the median | Scores low, but does not track engagement across your groups. Fixing it may matter for other reasons, but expect little change in engagement. |
| Maintain | Weak link, score above the median | Scoring well and not a lever. No action needed beyond keeping it there. |
In the benchmark map at the top of this page, communication, growth and development, involvement and recognition land in Fix first. Purpose alignment and meaningful work land in Protect. Rewards and work-life balance sit in Watch: low scores, but weaker levers for eNPS.
In a CultureMonkey leadership report, the What Moves Engagement block draws this chart from your own teams and names the strongest Fix first candidate in a sentence.
Is it a manager problem or an organization-wide one?
Before blaming managers for a low score, check whether it matters which manager someone has. The share of variation shows how much of the spread in engagement lies between groups, such as managers or locations, rather than within them.
- 14% or more
The grouping explains a lot of it. A manager split this large points to a management problem worth targeting.
- 6% to 14%
The grouping explains some of it. Worth looking at the weakest groups, alongside organization-wide action.
- Under 6%
The grouping barely matters here. The issue is organization-wide, and manager training alone is unlikely to fix it.
- They do not add up
Each grouping is a separate question. Manager and team can both explain a lot when they overlap.
This check stops a common mistake: launching a year of manager development for a problem that sits evenly across the whole organization. The What Explains Engagement block reports these bands for manager, team and location side by side.
Should the outcome be engagement or eNPS?
Use the outcome leadership cares about, and keep it consistent between surveys. Engagement and eNPS are related, but they often rank drivers differently, because what makes people engaged in their work is not always what makes them recommend the company.
If you use eNPS, compare drivers with the share of promoters rather than the eNPS score itself. eNPS is promoters minus detractors, so two very different groups can share one number; the share of advocates is the thing a driver can plausibly move. That is how the What Makes an Advocate block ranks drivers.
What if an average hides two camps?
An average of 7 out of 10 can be a contented team or two camps. Before ranking a driver, check whether its answers sit at both ends of the scale, roughly the bottom 40% and the top 20%, rather than in the middle.
A split driver needs a different response from a low one. A team that disagrees with itself usually has two different experiences, by shift, site or manager, and the fix starts with finding which group is which. The Where Opinion Is Split block lists the most divided groups for you.
What are the common mistakes in driver analysis?
The common mistakes in driver analysis are ranking by lowest score, reading correlation as cause, running it on too few groups, breaking anonymity to get more groups, trying to fix every driver at once and ignoring the drivers that are already strong.
- Ranking by lowest score
The lowest-scoring driver is the most visible, not necessarily the most important. Rank by link to the outcome, then look at score.
- Reading correlation as cause
A driver that moves with engagement is a candidate to act on, not proof that acting on it will move engagement. Test the action with a pulse.
- Running it on too few groups
Five teams can make any driver look decisive. Below about eight groups with enough responses, report that it is too early to rank.
- Breaking anonymity to get more groups
Cutting by smaller and smaller groups to get more data points exposes individuals. Every group in the analysis must clear the anonymity threshold.
- Fixing everything at once
A list of ten Fix first drivers is a wish list. Pick one or two, name an owner and check them at the next pulse.
- Ignoring the Protect zone
Strong drivers erode when they are taken for granted. Keep at least one Protect driver on the leadership agenda.
How do you present driver analysis to leadership?
Show driver analysis to leadership as one chart and one decision. The chart is score against link to the outcome, with the four zones marked. The decision is the one or two Fix first drivers you propose for the year, each with an owner and the survey question that will show whether it worked.
It belongs on the priority drivers slide of the results presentation, and as the "what moves engagement" part of the leadership report. Mark which changes since last year are real before you show any trend, as covered in how to interpret survey results, and compare against the last survey so leaders see movement, not just a snapshot.
How CultureMonkey runs key driver analysis
In CultureMonkey, leadership reports run this analysis as built-in blocks. What Moves Engagement correlates each driver with engagement across your teams, places every driver in Protect, Fix first, Watch or Maintain and names the strongest candidate in a sentence. It needs at least eight groups and says so when there are fewer.
Three more blocks cover the checks above: What Explains Engagement for the share of variation by manager, team or location, What Makes an Advocate for the link to the share of promoters, and Where Opinion Is Split for divided groups. How to read each one is in the Help Center.
Conclusion
The lowest score is where a driver stands, not whether it matters. Key driver analysis adds the second question, how closely each driver moves with engagement, and points the year's effort at the drivers that both score low and move the outcome.
Run it on enough groups, keep every group above the anonymity threshold, check whether the problem sits with managers or the whole organization, and pick one or two drivers rather than ten. Then test the action with a pulse before calling it a win.
Frequently asked questions
What is key driver analysis in an employee engagement survey?
Key driver analysis is a way of ranking engagement drivers, such as communication, growth and recognition, by how closely they move with an outcome like the engagement score or eNPS across teams. It shows which drivers are levers for engagement, rather than simply which ones score lowest.
Is the lowest-scoring driver always the one to fix?
No. In CultureMonkey's benchmark data, the two drivers most closely linked to eNPS, communication and growth and development, rank 16th and 18th of 22 on score, while rewards scores lowest but has only a moderate link to eNPS. Fix the drivers that both score below your median and move engagement.
How many teams do you need for driver analysis?
You need enough groups for a correlation to be stable, typically at least eight teams or segments that each clear your anonymity threshold. With fewer groups, the ranking can change completely from one survey to the next, so it is better to report that it is too early to tell.
Does key driver analysis prove what causes engagement?
No. Key driver analysis shows association: drivers that move together with engagement across groups. It is a strong guide to where to act, but whether an action changes engagement should be checked with a follow-up pulse survey on that driver.
Should I use engagement or eNPS as the outcome?
Use the outcome leadership cares about and keep it consistent. Engagement and intent to recommend are related but often rank drivers differently, so a driver analysis against eNPS can point to different priorities than one against the engagement score.
What does an r of 0.5 mean in driver analysis?
An r, or correlation coefficient, runs from -1 to 1 and shows how closely two measures move together. In CultureMonkey's benchmarks, a link of 0.5 or more between a driver and eNPS is treated as strong, 0.35 to 0.5 as moderate and anything lower as weak.