Boards still gamble on one person at the top, then scramble when succession turns into a cliff edge. The work underneath that role has changed. Teams coordinate across time zones, software cycles, and shifting priorities. In this environment, concentrating authority in a single pair of hands creates fragility. The alternative is not leaderless drift but leadership by rotation and design.
What the research actually shows
The best evidence comes from teams, where scholars can pool many studies and compare designs. A meta-analysis of 50 effect sizes across 3,198 teams found a positive relationship between shared leadership and performance, with stronger effects in real organizational settings than in classroom or lab studies (D'Innocenzo, Mathieu and Kukenberger, 2016). How researchers measured “shared” also mattered. Studies that used social network features such as density and (de)centralization reported larger effects than studies that simply aggregated individual ratings. That is a practical clue. When influence is genuinely distributed across many ties, outcomes improve more than when a nominally shared setup still revolves around a few voices.
A second independent meta-analysis across 42 samples estimated the overall correlation between shared leadership and team effectiveness at rho equal to 0.34, a moderate association (Wang, Waldman and Zhang, 2014). The content of what gets shared matters. Sharing so-called new-genre leadership behaviors such as transformational or charismatic leadership correlated more strongly with effectiveness, around rho equal to 0.34, than sharing traditional task and consideration behaviors at around rho equal to 0.18. The uplift was larger when the team’s work was more complex, a condition that fits knowledge work and self-managed settings. The authors also reported unique, incremental effects of shared leadership beyond what formal vertical leadership explained.
These are correlations, not guarantees. They do not claim that any rotation will work in any context. They do show that when leadership functions are spread across a web of peers, teams perform better on average, especially where the work resists simple scripts. For executives, the implication is to design for influence you can see and test, not for titles you hope will cascade.
From teams to the top: what co-CEOs tell us
The leap from teams to the corner office requires caution. Meta-analytic findings sit at the team level. Extending them to co-CEOs or rotating chief roles is an analogy, not a direct causal bridge. That said, the recent record suggests shared leadership at the top is more workable than folklore allows.
A Harvard Business Review assessment studied 87 companies led by co-CEOs and concluded that the model is rarer than myth but more viable than many assume (Feigen, Jenkins and Warendh, 2022). Among roughly 2,200 companies in the S&P 1200 and Russell 1000 between 1996 and 2020, fewer than 100 used co-chief executives, which underscores how unusual the arrangement remains. The authors point to failures, including Chipotle Mexican Grill, SAP, and Research In Motion/BlackBerry, especially under stress. The central argument is not that co-CEOs outperform, but that with clear domains, governance, and conflict rules, two heads at the top can function.
Take the three strands together. Shared leadership in teams is positively associated with better outcomes, and the effects are stronger when influence is widely and visibly distributed rather than merely declared (D'Innocenzo, Mathieu and Kukenberger, 2016). The gains are larger as complexity rises, and the content of leadership that gets shared matters, with new-genre behaviors linked to stronger effects (Wang, Waldman and Zhang, 2014). At the apex, co-CEO arrangements are rare but not inherently doomed, provided design and discipline exist (Feigen, Jenkins and Warendh, 2022). This does not certify rotating CEOs. It does make the case for structured rotation and succession as pragmatic options rather than last-resort fixes.
How to rotate leadership without losing the plot
The research points to two design levers: make influence legible, then decide what to rotate.
-
Map the network before you rotate. The first meta-analysis shows that social-network-based approaches aligned with stronger shared-leadership relationships to performance (D'Innocenzo, Mathieu and Kukenberger, 2016). Before any rotation, run a light-touch network inquiry. Ask who people go to for decisions, advice, and energy. Use that map to see whether authority and influence coincide. If your “succession bench” is invisible in the network, rotation will only shuffle titles.
-
Rotate leadership content, not just titles. The second meta-analysis highlights that new-genre leadership behaviors, such as vision-setting and inspiration, carry distinct weight when shared (Wang, Waldman and Zhang, 2014). Design rotations that make people explicitly responsible for different leadership functions: one cycle focused on strategic framing, another on culture and cohesion, a third on execution cadence. Treat each as a defined job with decision rights, peers, and metrics. Rotating these functions builds range and tests depth.
-
Keep domains and escalation rules crisp. The co-CEO record underscores the importance of clear swim lanes and conflict mechanisms when power is shared at the top (Feigen, Jenkins and Warendh, 2022). Apply the same logic to any rotation. Publish who decides, who is consulted, and how ties get broken. Ambiguity is the enemy of speed.
-
Match rotations to complexity. Shared leadership’s association with effectiveness strengthens as work gets more complex (Wang, Waldman and Zhang, 2014). Start rotational experiments in product, platform, or client units where interdependence is high. Do not start in routine units that run on standard work.
A practical cadence for succession
Treat rotational leadership as a continuous audit of your pipeline.
-
Start with a six to twelve month pilot inside one complex team. Measure baseline outcomes and the leadership network before the pilot. Run mid-cycle and end-cycle check-ins to see whether influence is spreading or collapsing back to a few nodes. The goal is to strengthen network density appropriately, not to dilute accountability. This approach aligns with the evidence that real-world, network-aware shared leadership aligns with better performance (D'Innocenzo, Mathieu and Kukenberger, 2016).
-
Add a co-lead bridge for critical transitions. Where risk is high, pair an incoming lead with a standing counterpart with non-overlapping domains. Borrow the co-CEO lesson on domain clarity and pre-agreed escalation, then time-box the pairing to avoid permanent duality unless the role genuinely merits it (Feigen, Jenkins and Warendh, 2022).
-
Use rotations to test for new-genre leadership. Assign cycles that require transformational behaviors, then evaluate with 360 input focused on those behaviors. The 2014 meta-analysis suggests that sharing these behaviors links more strongly to team effectiveness than rotating purely task-structuring duties (Wang, Waldman and Zhang, 2014).
-
Close the loop with visible criteria. Publish the evidence you track and the decisions you make from it. People will accept rotation when they can see the rules and the results.
The payoff is subtle but material. You are not trying to produce leaders faster. You are trying to distribute leadership work so that more people can carry it, and so that succession becomes routine. The evidence does not promise that rotation beats every alternative. It does show that when leadership is a networked practice, teams tend to perform better, especially under complexity. Combine that with disciplined co-lead experiments at the top, and you can replace hero bets with a system that learns.
Sources
- D'Innocenzo, Mathieu and Kukenberger (2016). A Meta-Analysis of Different Forms of Shared Leadership–Team Performance Relations. Finds a positive association between shared leadership and team performance, stronger with social-network measures and in field settings. https://journals.sagepub.com/doi/10.1177/0149206314525205
- Wang, Waldman and Zhang (2014). A meta-analysis of shared leadership and team effectiveness. Estimates an overall rho of 0.34 for shared leadership and team effectiveness, stronger in complex work and with unique effects beyond vertical leadership; new-genre behaviors matter most. https://pubmed.ncbi.nlm.nih.gov/24188392/
- Feigen, Jenkins and Warendh (2022). Is It Time to Consider Co-CEOs? Examines 87 co-CEO companies, noting the model is rare, has notable failures, yet can be viable with clear domains and governance. https://hbr.org/2022/07/is-it-time-to-consider-co-ceos