Boards keep searching for a single savior. Then a crisis hits, the lone chief falters, and succession looks like a cliff rather than a handover. There is a different bet: distribute authority ahead of time, rotate who holds the pen, and make leadership a team sport. The cleanest data sits below the C-suite, yet its signal is hard to ignore.
What the evidence actually says about shared leadership
Across 3,198 teams, a meta-analysis found a positive relationship between shared leadership and performance, with stronger effects in field settings than in labs and classrooms (D'Innocenzo, Mathieu and Kukenberger, 2016). How researchers measured sharing also mattered. Studies that mapped influence as a network, focusing on density and decentralization, reported larger effects than those that simply aggregated individual ratings. In short, outcomes improve when leadership truly circulates among people, not when a team retrofits a single leader’s style into a group average.
A second, independent meta-analysis of 42 samples reached a similar conclusion and put a number on it: shared leadership correlated with team effectiveness at roughly rho = .34 (Wang, Waldman and Zhang, 2014). The pattern was not uniform. Sharing “new-genre” leadership behaviors such as transformational or charismatic influence was linked more strongly to effectiveness than sharing classic initiating-structure or consideration behaviors. Most importantly for knowledge work, the positive effects were stronger when team tasks were more complex (Wang, Waldman and Zhang, 2014). If your work requires synthesis, judgment, and interdependence, distributing the lead seems to help.
These are team-level findings, not CEO trials. Extending them to the top job is an analogy, not a proof. Still, both meta-analyses point in the same direction. When influence flows through a network rather than sitting with one node, performance tends to rise, particularly in real organizations and on complex work (D'Innocenzo, Mathieu and Kukenberger, 2016; Wang, Waldman and Zhang, 2014). That is the empirical spine any discussion of rotating leadership and succession should start from.
What this implies for the top of the house
Evidence on shared leadership among top executives is thinner, but there is a relevant contemporary datapoint. A study of 87 companies with co-CEOs argues the model is more viable than is commonly assumed, even if rare. Fewer than 100 of roughly 2,200 firms in the S&P 1200 and Russell 1000 between 1996 and 2020 used co-chiefs (Feigen, Jenkins and Warendh, 2022). The record is mixed, as critics point out. The same article notes notable failures under stress, including Chipotle Mexican Grill, SAP, and Research In Motion/BlackBerry. The important takeaway for boards is not that co-CEO structures guarantee better performance. It is that shared top leadership can work, it remains unusual, and its success likely turns on design and context rather than the label alone (Feigen, Jenkins and Warendh, 2022).
Marrying the two literatures calls for humility. The team meta-analyses provide strong correlational evidence. They do not specify the best cadence for rotating a CEO or the optimal split of investor relations and operations between co-chiefs. Yet they do offer two practical anchors. First, the more complex the work, the more a shared model appears to help (Wang, Waldman and Zhang, 2014). Second, the way leadership is actually distributed matters more than the formal structure. Network reality beats org chart fiction (D'Innocenzo, Mathieu and Kukenberger, 2016).
Designing rotation: from principle to practice
Treat rotating leadership as an operating hypothesis, not a creed. Start where the upside is most plausible: units with high interdependence and cognitive complexity. The evidence says shared leadership pays off more in such contexts (Wang, Waldman and Zhang, 2014). Pair that with a hard-nosed audit of how influence really flows. Social-network measures of leadership, such as mapping who others seek for guidance and decisions, were the strongest predictors of performance in the meta-analysis (D'Innocenzo, Mathieu and Kukenberger, 2016). Use that idea. Before announcing a rotation, surface the informal leaders already carrying decisive weight.
Then design the rotation so substance beats symbolism:
- Define domains, not personalities. If one leader holds product and the other capital markets, say so. The co-CEO study flags that shared tops can work, but some firms floundered in stress. Role ambiguity is a plausible culprit, so reduce it upfront (Feigen, Jenkins and Warendh, 2022).
- Time-box the experiment. Use a fixed cycle for rotation or co-leadership with a pre-agreed review gate. The meta-analytic evidence is correlational, so boards should learn their way in rather than lock in a structure.
- Measure the network. Re-run influence and decision-flow assessments midstream. The 2016 meta-analysis found network-based indicators aligned with stronger performance effects. Build that lens into governance so you can see whether leadership is genuinely shared or simply split in name (D'Innocenzo, Mathieu and Kukenberger, 2016).
- Match model to complexity. Push rotating or co-leadership toward complex, interdependent work where the evidence suggests larger gains, and keep highly standardized work under clearer single-point authority (Wang, Waldman and Zhang, 2014).
Guardrails for succession under stress
The co-CEO record includes prominent failures, especially under pressure, which is exactly when succession structures are tested (Feigen, Jenkins and Warendh, 2022). Put stress protocols in writing before you need them. Decide who has the casting vote on liquidity events, regulatory crises, or sudden talent departures. Clarify external representation. Investors, employees, and customers need to know who speaks for the firm on which issues and when. These safeguards do not contradict shared leadership. They make it legible.
Finally, be honest about what the science does and does not say. At the team level, shared leadership correlates with stronger performance, particularly in complex work, and network reality matters more than labels (D'Innocenzo, Mathieu and Kukenberger, 2016; Wang, Waldman and Zhang, 2014). At the top, shared leadership is rare but not fanciful. There are notable failures, yet a review of 87 co-CEO cases argues the model is more workable than its reputation implies (Feigen, Jenkins and Warendh, 2022).
The practical implication is clear. Stop waiting for a perfect heir and start building a system that does not depend on one. Pilot rotating or co-leadership where your work is most complex, measure the network you actually have, and install crisis rules before the storm. Succession will look less like a cliff and more like a relay.
Sources
- D'Innocenzo, Mathieu and Kukenberger (2016). A Meta-Analysis of Different Forms of Shared Leadership–Team Performance Relations, Journal of Management. Finds a positive shared leadership–performance relationship, stronger in field settings and when measured through influence networks, grounding the case for genuine distribution of authority. https://journals.sagepub.com/doi/10.1177/0149206314525205
- Wang, Waldman and Zhang (2014). A meta-analysis of shared leadership and team effectiveness, Journal of Applied Psychology. Reports an overall rho ≈ .34 and stronger effects in complex work, supporting the use of shared or rotating leadership in knowledge-intensive contexts. https://pubmed.ncbi.nlm.nih.gov/24188392/
- Feigen, Jenkins and Warendh (2022). Is It Time to Consider Co-CEOs?, Harvard Business Review. Reviews 87 co-CEO cases, shows the model is rare yet workable, and notes prominent failures under stress, informing design and guardrails for shared top roles. https://hbr.org/2022/07/is-it-time-to-consider-co-ceos