What drives team engagement: a manager's own behavior

Most organizations have a theory of what makes a good manager, and most of those theories lean on tenure. The manager who has been at the company a decade is assumed to run a more engaged team than the one who started three months ago. Span of control is treated the same way: a manager with a large team is seen as more senior, more capable, more effective.

We tested both assumptions directly. Using a year of behavioral data from 633 managers across 60 companies, we ranked 12 manager behaviors and traits by how strongly each one separates high-engagement teams from low-engagement teams. The dependent variable is DEBI, the Dynamic Engagement Behavior Index, a per-manager score for how engaged that manager's team is.

The ranking is unambiguous. The top predictors are all things a manager personally does: checking in, replying to feedback, recognizing people, and reporting their own well-being. Tenure has essentially no relationship with team engagement (d=0.07). Team size has a small negative one (d=−0.31). Experience and headcount are not what move the number.

The participation divide

Before reading the effect sizes, one pattern needs to be on the table. We split the 633 managers into DEBI quartiles. The bottom quartile (Q1, n=159) has a mean DEBI of 0.0 and near-zero values on almost every variable we measured.

These are not "bad managers" in the traditional sense. They are absent managers — people who exist in the system but are not using the platform. The DEBI calculation reflects that: no inputs, no engagement score.

So the effect sizes below overstate the difference between good and bad management. What they actually measure is the difference between participating and not participating. That is still a meaningful finding, but it is a different one, and it shapes how the rest of this study should be read. The real management question lives in the middle: among managers who do participate, what separates good from great?

Finding 1 — The ranking: behavior beats tenure and team size

We computed Cohen's d for each variable, comparing the top DEBI quartile against the bottom. The result is a clean ordering: every top driver is a behavior, and the two structural variables most managers assume matter — tenure and team size — sit at the bottom.

The top five drivers

1. Manager happiness (d=3.75). A manager's own reported well-being is the single strongest differentiator. Top-quartile managers average 3.74 on the inverted happiness scale (where 5 = very happy); bottom-quartile managers average 0.13. This variable partly reflects participation — you have to check in to report happiness — but among active managers, happier ones still produce more engaged teams.

2. Reply rate (d=3.43). Top-quartile managers reply to 88.6% of the feedback they receive. Bottom-quartile managers reply to 2.5%. Responding to feedback signals that an employee's voice was heard. Prior research in this program found managers with high reply rates produce team DEBI of 57.6 against 29.3 for non-repliers; this study reinforces that pattern.

3. Reply quality (d=2.66). When top-quartile managers do reply, their responses score 63.9 on the quality index, which measures relevance, positivity, empathy, and clarity. How a manager replies matters almost as much as whether they reply.

4. Check-in frequency (d=2.09). Managers who check in on at least a quarter of eligible days post DEBI scores roughly 10x higher than those who never check in. The relationship is monotonic: every incremental increase in check-in frequency corresponds to higher team engagement. The next section breaks this dose-response curve down in detail.

5. Recognition activity (d=1.36–1.39). Both giving and receiving recognition show large effects. Top-quartile managers give 31 recognitions per year against 0.5 for the bottom quartile. Receiving recognition (d=1.39) slightly edges out giving (d=1.36), which suggests that managers embedded in the recognition culture — on both sides of it — build more engaged teams.

The variables that do not matter

Tenure (d=0.07). How long a manager has been at the company has essentially zero relationship with team engagement. A manager with three months of tenure can produce the same engagement as one with ten years. This contradicts the common assumption that experience equals effectiveness.

Team size (d=−0.31). The one negative effect in the study. High-engagement managers have slightly smaller teams: 5.1 direct reports against 6.6 for the bottom quartile. The effect is small, but it points the same direction as research on span of control, where smaller teams allow for more personalized management. The team-size result is explored further below.

Finding 2 — The check-in dose-response

Check-in frequency produces the clearest monotonic dose-response pattern in the study. We binned managers by the share of eligible days they checked in and looked at mean team DEBI in each bin.

The biggest jump happens between 0% and 1–25%. A manager who checks in just once or twice a week already produces 10x the team engagement of one who never checks in — 33.0 against 3.4. Across the full range, from 0% to the 51–75% bin, mean DEBI rises from 3.4 to 51.9, a 15x gap. Returns diminish after the first step, but the curve keeps climbing.

Recognition follows a smoother curve

Recognition shows a similar relationship to check-ins, but more gradual. We binned managers by recognitions given over the year.

Managers who give 51+ recognitions per year reach the highest team engagement at DEBI 61.0. But even 1–5 recognitions per year triples the baseline, from 8.8 to 30.4. The curve is smooth and monotonic, and returns continue all the way to the top bin.

Finding 3 — The team-size paradox

One result runs against conventional wisdom: managers with larger teams produce lower engagement, not higher. Team size is the only variable in the study with a negative Cohen's d (−0.31) and a negative Pearson correlation with DEBI (r = −0.04).

The intuition might run: a manager with more reports has more people contributing to engagement metrics, so their score should be higher. The data says the opposite. The likely reason is that attention is finite. A manager with 5 direct reports can reply to feedback, give recognition, and check in on each person regularly. A manager with 10 reports has half the bandwidth per person, and the dose-response data shows engagement responds to frequency of contact. Frequency drops as the denominator grows.

The two measures diverge, and that divergence is informative. The Cohen's d of −0.31 compares extremes: top-quartile managers average 5.1 reports, bottom-quartile 6.6, a real gap of 1.5 people. But the Pearson r across all 633 managers is essentially zero (−0.04). Team size does not predict engagement in a linear way across the full range. It acts more like a ceiling: past a threshold, the odds of sustaining high engagement drop, but below that threshold, size alone guarantees nothing.

The practical read: keep spans of control under 6–7 where possible, but do not expect that shrinking a team will, by itself, raise engagement. What matters more — by a factor of 10x in effect size — is what the manager does with the team they have. Team size is a constraint, not a cause.

The correlation picture

Effect sizes compare quartile extremes. Pearson correlations across the full sample tell a consistent story: the behavioral variables correlate most strongly with DEBI, and tenure and team size do not.

What this means for HR

The ranking points to a small set of decisions that should change how organizations build and support managers. Each one follows directly from the data.

The low-engagement managers in this study are not doing harmful things. They are doing almost nothing on the platform. That reframes the intervention question. The goal is not "make bad managers better." It is "get inactive managers to participate at all" — and even minimal participation produces dramatic engagement gains.

Limitations

• Participation confound. The largest effects are driven by the participation divide, not by behavioral differences among active managers. A follow-up should filter to managers with meaningful activity (for example, check-in rate > 10%) and re-run the analysis.
• Cross-sectional design. The data cannot establish causality. Does manager happiness drive team engagement, or do engaged teams make managers happier? Likely both, in a reinforcing cycle.
• DEBI at zero. 25% of managers have DEBI = 0, which may reflect data gaps rather than truly zero engagement. The DEBI calculation may require minimum activity thresholds to produce a score.
• Power-skill multicollinearity. The six power skills are too correlated to isolate individual effects. They collectively predict DEBI (r = 0.37–0.51) but function as a single latent factor.
• Sparse performance data. Only 11.4% of managers have performance-rating data and 12.6% have boost data. Effect sizes for those variables should be read cautiously.