Shared Leadership: S-Tier Behavioral Designer’s Guide

Watch a great jazz quartet and try to find the leader. For eight bars it is the bassist holding everyone to the floor. Then the pianist takes a turn and the others fall in behind her. Then the drummer pushes the tempo and the whole group leans into it, then hands it back. Nobody appointed a conductor. The baton moves to whoever has the next thing the music needs, and the moment they have said it, they pass it on. The group is being led the entire time. It is just never being led by the same person twice in a row.

That is the picture Craig Pearce and Jay Conger were chasing when they pulled together Shared Leadership in 2003. For a century, leadership research had a built-in assumption so deep that almost nobody questioned it: leadership is a thing one person does to a group. Find the leader, study the leader, train the leader, and the team’s fate follows. Pearce and Conger asked the heretical question. What if leadership is not a role sitting at the top of a team but an influence that moves through a team, passing from member to member as the work demands? What if the unit of analysis is not the leader at all, but the pattern of who is leading whom, when, and about what?

Most organizations still get this wrong in the same expensive way. They try to “empower the team” by promoting a charismatic manager and telling everyone to take more ownership. Then they are baffled when ownership does not show up. The mistake is treating leadership as something you grant from the center, when shared leadership is a property of the network, not a gift from the boss. You cannot hand a team distributed influence the way you hand someone a title. You can only build the conditions in which it grows, and then get out of its way.

This guide rebuilds shared leadership from first principles. It covers the shift from vertical to lateral influence that defines the model, the enabling conditions that let leadership distribute instead of fragment, the way it emerges and how researchers actually measure it, the meta-analytic evidence on when it beats a single strong leader and when it does not, the sharpest criticisms of the idea, and the part no organizational-behavior textbook will hand you: a translation of distributed leadership into the eight Core Drives of the Octalysis Framework, so you can stop decreeing “everyone is a leader here” and start engineering the network conditions that make it true.

Speed Run Notes

• Shared leadership (Pearce & Conger, 2003) reframes leadership as a dynamic influence process that moves through a team, member to member, rather than a role one appointed person holds over it. The unit of analysis is the network, not the boss.
• It is not the absence of a leader. The vertical leader’s job inverts: from being the source of influence to building the conditions where peer influence can grow, then designing themselves out of the center.
• It needs three enabling conditions (Carson, Tesluk & Marrone, 2007): shared purpose, social support, and voice. Without them, distributing authority produces paralysis or power struggle, not empowerment.
• The evidence is real but conditional. Meta-analyses (Wang et al. 2014; D’Innocenzo et al. 2016) find shared leadership predicts team effectiveness above and beyond vertical leadership, most strongly on complex, interdependent knowledge work.
• The crosswalk: shared leadership is a Core Drive 4 (Ownership) and Core Drive 5 (Relatedness) network property, gated by Core Drive 1 (shared purpose) and Core Drive 3 (voice). Its failure mode is Core Drive 8 accountability diffusion: when everyone leads, no one feels responsible.
• The S-tier move is to build the enabling conditions first, then distribute authority, and to read a flat org chart with low measured influence density as a warning light, not a win.

Author Credibility: Yu-kai Chou

Yu-kai Chou created the Octalysis Framework after studying gamification since 2003 — years before the term entered mainstream vocabulary. As a Human-Systems Architect & Behavioral Designer, his framework has been applied by LEGO, Microsoft, Porsche, Coca-Cola, Salesforce, and MrBeast, impacting over 1.5 Billion Users.

Chou has taught the Octalysis methodology at Harvard, Stanford, Yale, Tesla, Google, BCG, and IDEO.

His work has been cited by Harvard, Stanford, MIT, Forbes, Wall Street Journal, Wired, US Department of Energy, NIST, NSF, NCBI, US Department of Education, ClinicalTrials.gov, and Google Scholar — with 3,700+ more academic publications. Explore his books here.

What Is Shared Leadership?

Shared leadership is a dynamic, interactive influence process among the members of a team, where the goal is to lead one another toward the team’s objectives, with leadership moving fluidly among members rather than residing in a single appointed person. Pearce and Conger formalized that definition in their 2003 edited volume, and the radical move is hidden in plain sight: leadership becomes a verb the whole team conjugates, not a noun one person owns. On any given afternoon, the most influential voice in a product team might be the engineer who understands the failing system, then the designer who sees the user’s confusion, then the analyst who has the data nobody else has read. Each one leads on the dimension where their expertise is decisive, and follows everywhere else.

That reframe broke from a tradition that ran almost unchallenged for a hundred years. From the Great Man theories of the early 1900s through trait research, behavioral models, and contingency theories, the leader was always the dependent variable’s cause and the team was the thing being acted upon. Even the relational models that softened the hierarchy, like leader-member exchange, kept the leader at the center of a hub-and-spoke wheel, trading different-quality relationships outward. Shared leadership cut the hub out of the picture. It asked researchers to map the whole web of influence among everyone, including the lateral ties between members that the vertical lens never sees.

The idea was not born in 2003. Cecil Gibb gestured at “distributed leadership” as early as 1954, and David Bowers and Stanley Seashore documented “mutual leadership” among peers in 1966, finding that the leadership members supplied to each other predicted effectiveness on top of what the formal supervisor supplied. The insight kept getting rediscovered and kept getting buried, because the dominant methods could only measure one leader at a time. What Pearce and Conger did was give the idea a name, a research program, and a measurement strategy precise enough that the field could finally study leadership as a team-level property instead of an individual trait.

The reframe forces a sharp distinction that practitioners constantly blur: shared leadership is not the absence of a leader, and it is not democracy. A leaderless team is not a shared-leadership team. It is usually a stuck team. The point is not that nobody leads but that everybody might, depending on the moment, and that the formal leader’s role shifts from supplying all the influence to cultivating the conditions where members can supply it to each other. Pearce was emphatic that this combines with vertical leadership rather than replacing it. The question is never “boss or no boss.” It is how much of the team’s total leadership comes from the top versus from the lateral web, and whether that mix fits the work.

From Vertical to Lateral: The Building Blocks

Before the evidence and the failure modes, the architecture. Shared leadership rests on three structural ideas, and skipping any of them is how the concept gets misused.

Vertical Leadership: The Default We Inherited

Vertical leadership is the model everyone defaults to: influence flows downhill from a formally appointed leader to followers. It is not wrong, and shared-leadership researchers never claimed it was. Vertical leadership is fast, it is clear, and it concentrates accountability in one place, which is exactly what you want when a decision has to be made in ten seconds and someone has to own the consequence. The problem is not that vertical leadership is bad. The problem is that it was treated as the only kind, so a single manager became the rate limiter on a team’s entire capacity to think. When the work outgrows what one mind can track, a purely vertical structure quietly caps the team at the bandwidth of its boss.

Shared Leadership as a Property of the Team

The conceptual leap is to treat leadership as something a team has in varying amounts and distributions, the way a network has density. In a vertical team, all the leadership ties point at one node. In a shared-leadership team, leadership ties run in many directions, member to member, and the team’s leadership is the sum and shape of that whole web. This is why the serious measurement uses social network analysis. You ask every member who they turn to for leadership on which dimensions, then look at the resulting network. A dense, distributed network means leadership is actually shared. A sparse one with a single hub means it is not, no matter what the org chart says.

Holding leadership as a team property rather than an individual one explains findings that the vertical lens cannot. Two teams can have identically rated managers and wildly different outcomes, because one team’s members lead each other and the other’s wait to be led. The difference lives in the lateral ties, which a leader-centric study never records. The whole methodological contribution of the 2003 program was making those invisible ties visible.

The Three Enabling Conditions

Shared leadership does not appear because someone wants it to. Jennifer Carson, Paul Tesluk, and Jennifer Marrone identified the soil it grows in, in their influential 2007 study of consulting teams: an internal team environment built from three ingredients. Shared purpose means members have a common, internalized understanding of what the team is for, which aligns the distributed influence so it pulls in one direction instead of scattering. Social support means members actively recognize and encourage each other’s contributions, which makes it safe to step up and lead. Voice means members have genuine input into how the team operates, so leading is something they are invited to do rather than something they have to seize.

Carson and colleagues added a second pathway: external coaching from a formal leader, which mattered most when the internal environment was weak. The two paths substitute for each other. A team rich in purpose, support, and voice generates shared leadership on its own. A team thin in those needs a coaching leader to scaffold it into existence. Notice what is absent from this list: a memo announcing flat hierarchy. The conditions are relational and slow to build, which is precisely why the shortcut of just removing managers reliably fails.

How Shared Leadership Emerges and How You Measure It

If leadership is a network rather than a person, then two questions become central that vertical models never had to ask: how does the network form, and how do you put a number on it?

The Claiming and Granting Cycle

Scott DeRue and Susan Ashford gave the cleanest account of how leadership distributes in real time. Leadership identities are not assigned. They are claimed and granted in a continuous social cycle. A member claims leadership by stepping into an influence attempt: framing the problem, proposing the path, taking initiative. The rest of the team grants it by following, deferring, treating that person as the leader for that moment. Claim and grant, and a leadership relationship forms. Claim and get no grant, and the attempt dies. Over hundreds of these micro-exchanges, a team works out a distributed, fluid pattern of who leads on what. Shared leadership is the residue of thousands of small claims that got granted across many different members.

This is why psychological safety is load-bearing. Claiming leadership is a risk: you might be wrong, you might overstep, you might get slapped down. Amy Edmondson’s research on team learning showed that members only take interpersonal risks when they believe the team will not punish them for it. In a low-safety team, people stop claiming, the lateral ties never form, and leadership collapses back to whoever has positional power. Safety is the precondition for the claiming half of the cycle to even happen.

Measuring Distribution: Density and Centralization

The breakthrough that let shared leadership become a real research object was borrowing two metrics from network science. Density measures how many leadership ties exist in the team relative to how many could exist: high density means many members lead many others. Centralization measures how concentrated those ties are: high centralization means one or two nodes dominate even if some sharing exists. A real shared-leadership team scores high on density and low on centralization. Ajay Mehra and colleagues showed in 2006 that the shape of the leadership network, not just its amount, predicts performance, and that some distributed patterns work far better than others.

This matters enormously for design, because it gives you a diagnostic that cuts through the rhetoric. An organization can declare itself flat, abolish titles, and still have a leadership network that is sparse and centralized, where one informal alpha runs everything and the “flatness” is theater. The density-and-centralization measurement catches that immediately. You do not ask whether the team feels empowered. You map who actually leads whom, and the network tells you the truth the survey cannot.

What Pearce and Conger Got Right

Three contributions of the shared-leadership program have held up across two decades and several meta-analyses, and they are worth naming precisely.

First, the dependent-variable flip was a real break. By making leadership a team-level network property rather than an individual’s behavior, the model could explain the gap between two teams with equally good managers. Wang, Waldman, and Zhang’s 2014 meta-analysis found that shared leadership related to team effectiveness more strongly than vertical leadership did, with the gap widening for attitudinal, cognitive, and behavioral outcomes and for complex tasks. The effect was not a rounding error on top of the boss. On the work that matters most in a knowledge economy, the lateral web did more than the hierarchy.

Second, the model correctly predicted its own boundary conditions. D’Innocenzo, Mathieu, and Kukenberger’s 2016 meta-analysis showed that the shared-leadership-to-performance link grew stronger as task interdependence rose and as teams matured. This is the model behaving like a real theory rather than a slogan: it told you in advance where it would and would not work. Shared leadership pays off when members genuinely need each other’s expertise and have had time to build the network. On routine, independent work, or in a team that formed last week, the predicted payoff is small, and the data agree.

Third, Pearce’s insistence on combining vertical and shared leadership saved the idea from its own zealots. The 2004 framing was that the vertical leader’s highest function in knowledge work is to develop shared leadership, not to perform all of it. Michelle Ensley and colleagues found in 2006 that new-venture top teams did best with both forms present, vertical leadership setting direction and shared leadership distributing the execution. The mature position is not “destroy hierarchy.” It is “use hierarchy to grow the lateral capacity, then lean on the lateral capacity for the work that exceeds any one person.”

Where Shared Leadership Falls Apart

A concept this fashionable attracts uncritical adoption, and the honest version of this guide has to give the failure modes their due. Three lines of criticism land hardest.

When Everyone Leads, No One Is Accountable

The deepest danger is accountability diffusion. The same psychology that produces the bystander effect, where a crowd watches an emergency and nobody acts because responsibility is smeared across everyone, can poison a shared-leadership team. Bibb Latané and colleagues documented social loafing for exactly this reason: as a group grows, each member’s felt responsibility shrinks, and effort drops. Distribute leadership without distributing clear ownership and you can manufacture a team where everyone assumes someone else is steering, and nobody is. “Shared” quietly becomes “abdicated.” This is not a fringe risk. It is the default outcome of removing structure without replacing it with the enabling conditions.

The Coordination Cost Is Real

Fluid influence is not free. Every time leadership has to move from one member to another, the team pays a coordination tax: who decides this, whose call overrides whose, how do we resolve it when two members both claim leadership on the same dimension? In a vertical team that question has a one-word answer. In a shared-leadership team it has to be renegotiated, and without strong shared purpose to align everyone, the renegotiation curdles into a status contest. The model’s benefits assume the coordination cost is lower than the bandwidth gain. On simple work where one person could easily track everything, the coordination cost dominates and shared leadership just makes a simple job slow.

Does It Cause Performance, or Just Follow It?

The sharpest methodological critique is endogeneity. Most shared-leadership studies are cross-sectional: they measure the network and the performance at the same time, then report a correlation. But high-performing teams may develop shared leadership precisely because they are succeeding, with success building the trust and safety that let members claim and grant. If that is the dominant direction, then telling a struggling team to “share leadership more” is prescribing the symptom of health as if it were the cause. The longitudinal work and the interdependence moderators push back on the pure-reverse-causation story, but any honest practitioner holds the possibility that shared leadership is often a marker of a good team rather than a lever that makes a team good.

What’s Really Happening Inside the Brain

The shift from being led to leading, and back again, is not just an organizational arrangement. It runs on specific neural systems, which is the deepest reason the model’s enabling conditions are not optional.

Leading activates the brain’s agency and reward machinery. When a person exercises genuine influence over an outcome, rather than merely complying with someone else’s call, the striatal reward response to that outcome is amplified. Lauren Leotti and Mauricio Delgado showed that the mere opportunity to exercise control is itself rewarding, lighting up the ventral striatum independent of any material payoff. The brain treats agency as intrinsically valuable. This is why distributed leadership, done right, energizes a team: every member who gets to lead on their dimension is metabolizing a reward signal that a permanent follower never receives. A team where only one person ever leads is a team where only one person’s agency circuitry is being fed.

Granting leadership runs on the social brain. When the team defers to a member’s influence attempt, that member experiences a status-and-belonging signal that Matthew Lieberman’s work locates in the same circuitry the brain uses for physical reward and physical pain. Being granted leadership is socially rewarding. Being ignored when you claim it registers as a form of social rejection, processed through the dorsal anterior cingulate. This is the neural reason psychological safety gates the whole system: in an unsafe team, the predicted cost of a failed claim (social pain) outweighs the predicted reward of a successful one, so the rational brain stops claiming, and the lateral network never forms.

Accountability diffusion has a signature too. Personal agency depends on a clear link between my action and the outcome, a link the medial prefrontal cortex tracks as self-relevance. When responsibility is smeared across many people, that self-other attribution blurs, the personal-agency signal weakens, and the felt obligation to act drops. The bystander effect is partly this blurring made visible. The design lesson the biology hands you is blunt: distributing the opportunity to lead feeds the reward and agency systems, but distributing accountability without preserving individual ownership starves the very signal that makes people step up. You want shared influence sitting on top of clear, individual ownership, not shared influence replacing it.

Shared Leadership vs Other Frameworks

Shared leadership did not arrive in a vacuum. Placing it next to its neighbors sharpens what is actually distinctive about it.

vs Transformational and Vertical Leadership

Transformational leadership, the dominant model of the era, is still vertical in structure: one inspiring leader transforms followers through vision, charisma, and individualized attention. Shared leadership does not reject it so much as relocate it. The question shifts from “how good is the leader” to “how well is leadership distributed,” and the most useful synthesis treats a transformational vertical leader as the gardener whose job is to grow the lateral network. The two are complementary layers, not competitors. A team can have a visionary boss and a dense peer-leadership web at once, and the evidence says that combination outperforms either alone.

vs Leader-Member Exchange

Leader-member exchange and shared leadership both move past the one-size-fits-all leader, but in opposite directions. Leader-member exchange keeps a single leader at the center and varies the quality of the relationships radiating out to each follower. Shared leadership removes the privileged center entirely and studies the full lateral web. One is a hub with spokes of different thickness. The other is a mesh. Reading them together is clarifying: leader-member exchange describes the vertical ties, shared leadership describes the lateral ones, and a complete picture of a team’s influence needs both maps overlaid.

vs Team Mental Models

Shared leadership and shared cognition are easy to confuse and importantly different. Team mental models are about shared understanding: the team holds compatible representations of the task and of each other, which lets them coordinate without constant talk. Shared leadership is about shared influence: who gets to steer. The two reinforce each other, because a team that understands each other’s expertise knows whom to grant leadership to in a given moment. Shared mental models tell the team who should lead now. Shared leadership is the team acting on that knowledge.

vs Self-Managing Teams

Self-managing teams are the organizational structure; shared leadership is the influence process that may or may not fill it. Giving a team formal authority over its own work creates the space for shared leadership but does not guarantee it. A self-managing team with low density and high centralization has the autonomy of self-management with the influence pattern of a hierarchy, run by one informal boss the org chart pretends does not exist. Structure is the container. Shared leadership is what does or does not grow inside it, which is why the network measurement matters more than the formal design.

Shared Leadership in the Real World

The model earns its reputation in the field, where the network lens repeatedly explains outcomes that leader-centric thinking gets wrong.

Knowledge Work and Software Teams

Knowledge work is shared leadership’s home turf, because expertise is distributed by definition and no manager can be the smartest person in the room on every dimension. A modern software team has the database expert, the front-end specialist, the security reviewer, and the product thinker, and the leadership has to flow to whoever owns the decisive expertise for the task in front of them. This is why practices like rotating tech leads, collective code ownership, and blameless retrospectives work: they are mechanisms for distributing the claiming-and-granting cycle. The teams that ship reliably are rarely the ones with the most brilliant lead. They are the ones where leadership moves cleanly to the right person and back.

Healthcare: The Operating Room and the Care Team

Healthcare shows both the promise and the precision the model demands. A surgical team needs vertical leadership at the decisive moment, the surgeon’s call is final when seconds count, and shared leadership everywhere else, because the anesthesiologist, the scrub nurse, and the perfusionist each lead on the dimension only they can see. The deadliest failures in operating rooms are often failures of a junior member not claiming leadership when they spotted a problem, because the team’s safety was too low to make the claim survivable. The shared-leadership prescription is not to flatten the surgeon’s authority in a crisis. It is to build the safety that lets every set of eyes lead on what it catches, precisely so the vertical authority has better information when it does decide.

Startups and Founding Teams

Founding teams are a natural experiment in shared leadership, and Ensley’s research on new ventures found that the blend matters. Early-stage companies face problems no single founder has the full skill set for, so the founder who insists on being the sole leader caps the venture at their own limits. The teams that scale tend to run dense shared leadership inside a light vertical frame: a clear ultimate decision-maker for the calls that need one, and genuine peer leadership across product, growth, and operations. The failure pattern is the founder who confuses being the boss with being the only person allowed to lead, and starves the venture of every other brain on the team.

Crisis Response: Where It Breaks

Crisis is the model’s honest counterexample. When a building is on fire, you do not want a dynamic, interactive influence process. You want one incident commander and instant compliance, because the coordination cost of distributing leadership is fatal when the clock is in seconds. This is not a flaw in shared leadership. It is the boundary condition the model itself predicts: low time, high stakes, and the need for one accountable decision all push toward vertical leadership. The skilled move is matching the influence structure to the moment, dense and shared for the complex design work, sharp and vertical for the emergency, and knowing which one you are in.

The Elephant in the Room

Here is what most “flat organization” announcements will not say out loud: removing managers is cheap and politically flattering, while building the conditions for real shared leadership is expensive and slow, so companies do the first and call it the second.

Abolishing titles, deleting a management layer, and telling everyone they are now empowered costs nothing and reads as bold and modern. Building genuine shared leadership requires psychological safety that takes months to earn, a shared purpose people actually internalize, real channels for voice, and coaching to scaffold the network until it stands on its own. One of these can be done by Friday with a slide deck. The other is years of patient relational work that implicates how leaders behave every single day. So the slide deck wins, the managers disappear, the enabling conditions never get built, and the predictable result is a team that is not liberated but rudderless, where the loudest person quietly becomes the unaccountable boss and the org chart’s “flatness” hides a more arbitrary hierarchy than the one it replaced.

The deeper dodge is using “shared leadership” to offload the work of leadership without offloading the authority or the rewards that should come with it. Tell people they are all leaders now, expect them to take on the coordination, the conflict resolution, and the accountability of leadership, but keep the decision rights and the compensation concentrated at the top, and you have not distributed leadership. You have distributed unpaid labor and called it empowerment. The honest test borrows the publicity principle from libertarian paternalism: would you tell the team plainly that you are asking them to absorb the burdens of leadership without the corresponding authority, voice, or reward? If that sentence is one you would not say to their faces, the design is extraction wearing the costume of distribution, and the defensible version is the conjunction: distribute the influence and the decision rights and the recognition together, or do not call it shared leadership.

How to Apply Shared Leadership with Octalysis

Shared leadership tells you that influence can distribute across a team and that it pays off on complex, interdependent work. What it does not tell you is why a person steps up to lead, or which motivational levers make the claiming-and-granting cycle run instead of stall. That is the gap the Octalysis Framework fills. Octalysis decomposes all human motivation into eight Core Drives, and once you map the enabling conditions and failure modes of shared leadership onto those Core Drives, “everyone is a leader here” stops being a poster and becomes a design problem with a specific blueprint.

The Shared Leadership × Octalysis Core Drive Crosswalk

Shared leadership is not one motivation. It is a specific stack of Core Drives, and naming the stack is what lets you stop asking “is the team empowered?” and start asking “which Core Drive is under-supplied in the influence network I am trying to grow?”

The headline driver is Core Drive 4 (CD4): Ownership & Possession. When you are allowed to lead on something, you own it, and ownership is the engine that turns a passive team member into someone who steers. Distributed leadership is distributed ownership: each member feels that the team’s outcome is partly theirs to shape, not the boss’s alone. This is the Core Drive the entire Octalysis Framework is named after, and it is the deepest reason shared leadership energizes rather than exhausts. A permanent follower owns nothing and protects nothing. A member who leads on their dimension owns that dimension and defends it.

The medium it travels through is Core Drive 5 (CD5): Social Influence & Relatedness. Shared leadership is, by definition, a peer-to-peer influence process, which is pure CD5. The claiming-and-granting cycle is members influencing members; Carson’s “social support” condition is CD5 made into a design requirement; mentorship, peer accountability, and the felt pull of letting your teammates down are all CD5 mechanics. Where vertical leadership runs influence down a chain of command, shared leadership runs it across a web of relationships, and the strength of that web is a CD5 measurement.

The permission to lead is Core Drive 3 (CD3): Empowerment of Creativity & Feedback. Carson’s “voice” condition is CD3 by another name: members exercise real judgment, make meaningful choices about how the work goes, and see their input change the outcome. Without CD3, there is nothing to lead with. A member who cannot make a consequential decision cannot claim leadership, because there is no decision to claim. Voice is the raw material the claiming half of the cycle works on.

The alignment that keeps it from fragmenting is Core Drive 1 (CD1): Epic Meaning & Calling. Carson’s “shared purpose” condition is CD1. This is the Core Drive that turns distributed influence into a team pulling one direction rather than a committee pulling apart. When everyone leads and there is no shared purpose, you get the coordination tax and the status contest. When everyone leads under a purpose they have internalized, the distributed influence self-aligns, because each member is steering toward the same horizon. CD1 is the difference between a jazz quartet and a traffic jam.

The merit signal that routes leadership is Core Drive 2 (CD2): Development & Accomplishment. Leadership in a healthy shared system flows to whoever has the task-relevant expertise at the moment, which makes it a dynamic CD2 meritocracy: you lead because you have demonstrably mastered the dimension at hand, and the grant is the team’s recognition of that mastery. This is what keeps shared leadership from collapsing into pure politics. The grant should track competence, not volume.

The failure mode is Black Hat run amok. The accountability-diffusion trap runs on Core Drive 8 (CD8): Loss & Avoidance: when leadership is shared but ownership is not, the fear of being blamed diffuses across the group, and each member avoids stepping up because someone else might, which is the bystander effect in Core Drive terms. The status contest that erupts without shared purpose runs on Core Drive 6 (CD6): Scarcity & Impatience, as members compete over a perceived scarce supply of influence. Unrelieved CD8 and CD6 are the signature of shared leadership gone wrong: lots of activation, no fulfillment, and a team that is either paralyzed or at war.

The S-Tier Inversion: Build the Conditions, Then Distribute Authority

Here is the unique-to-Yu-kai move the crosswalk makes obvious. Because shared leadership is a CD4 and CD5 network property gated by CD1 and CD3, it cannot be decreed. You cannot grant a team distributed influence the way you grant a raise. You can only build the enabling conditions and let the network emerge. So the order of operations is fixed and almost everyone gets it backward: build the conditions first, then distribute the authority.

The default move is to distribute authority first, delete the managers, flatten the chart, announce that everyone leads, and hope ownership shows up. The crosswalk says it will not, because distributing authority before the CD1 purpose, CD3 voice, and CD5 safety exist drops the team into one of two Black Hat traps. With no shared purpose, the freed influence becomes a CD6 status contest. With no safety, the freed influence triggers CD8 diffusion and nobody claims it. Either way you get the rudderless team the elephant section described. The conditions are not a nice-to-have you add after going flat. They are the precondition that makes flat survivable.

So the S-tier sequence inverts the usual one. First build CD1 shared purpose so distributed influence has something to align to. Then build CD3 voice and the psychological safety that makes claiming leadership survivable. Then build CD5 social support so granting becomes the team’s habit. Only then distribute the formal authority, and do it on top of preserved individual ownership so the CD8 diffusion trap never opens. And the dashboard rule that follows: read a flat org chart sitting next to a low measured influence density as a warning light, not a win. A team that calls itself flat but whose leadership network is sparse and centralized has not distributed leadership. It has hidden its hierarchy. The diagnostic is the network, never the announcement.

The 6-Step Shared Leadership × Octalysis Audit

To make the crosswalk operational, run the team through six steps rather than guessing.

1. Map the influence network. Ask every member who they turn to for leadership on which dimensions. Compute density and centralization. This is your ground truth, not the org chart.
2. Score the four enabling conditions as Core Drives. Rate CD1 shared purpose, CD3 voice, CD5 social support, and the psychological safety that gates them. A weak score on any one predicts where the network will fail to form.
3. Check ownership is individual, not diffused. For each major outcome, confirm one member owns it even though influence is shared. Shared influence on top of clear ownership is healthy; shared influence replacing ownership is the CD8 diffusion trap.
4. Run the publicity test. For any planned change, ask whether you would tell the team plainly that you are distributing the burdens of leadership along with the authority, voice, and recognition. If you are distributing only the burdens, it is extraction, not sharing.
5. Build conditions, then distribute authority, in that order. Strengthen the weakest enabling condition before flattening anything. Distributing authority into thin conditions opens a CD6 or CD8 trap.
6. Re-map at 6 and 18 months. The 18-month re-map is the durability gate. It catches the flat-org announcement that spiked enthusiasm for a quarter before the network quietly re-centralized around one informal alpha.

Mapping to the Four Experience Phases

Octalysis tracks motivation across four phases of a person’s journey, and the balance of vertical and shared leadership has to shift across them. In Discovery, a new team needs more vertical leadership and high CD1 purpose, because the network has not formed yet and a clear direction prevents early chaos. In Onboarding, deliberately seed the claiming-and-granting cycle: assign small, safe leadership opportunities so members practice claiming with CD3 voice and the team practices granting with CD5 support, building safety through low-stakes reps. In Scaffolding, the long mid-tenure stretch, the influence network should densify and decentralize as expertise becomes visible and CD2 competence routes leadership to the right member, while you watch for the CD6 status contests that emerge as the stakes rise. In Endgame, the durable state is a dense, low-centralization network held together by deeply internalized CD1 purpose, with the vertical leader having largely designed themselves out of the day-to-day flow, present mainly to protect the conditions and make the rare call that needs one accountable owner.

Practical Steps for Building Distributed Influence

If you lead a team, or sit inside one, here is the model turned into action.

1. Map who actually leads whom. Before changing anything, ask the team privately who they turn to for direction on which kinds of problems. The pattern that comes back is almost never the org chart, and it is your real starting point.
2. Fix the weakest enabling condition first. Of shared purpose, voice, and social support, find the thinnest and strengthen it before touching structure. Distributing authority into a team without these manufactures chaos, not ownership.
3. Make claiming leadership safe. The single highest-leverage move is psychological safety. Respond to a member’s first risky influence attempt with curiosity, not correction, and the whole claiming cycle opens up. Slap down one claim and you teach the team to stop claiming.
4. Route leadership to expertise, not to volume. Explicitly grant leadership to the member who owns the relevant expertise for the task at hand, even when they are junior or quiet. This makes the grant track CD2 competence instead of CD6 dominance.
5. Preserve individual ownership underneath shared influence. For every important outcome, name one owner even as the influence stays distributed. This is how you get the energy of shared leadership without the CD8 accountability-diffusion trap.
6. Shrink your own footprint on purpose. If you are the formal leader, your success metric is how little the team needs you in the room. Measure the days the team made good calls without you, and treat a rising number as the goal, not a threat.
7. Match the structure to the moment. Run dense and shared for complex, interdependent design work. Snap to sharp and vertical for the crisis that needs one fast, accountable decision. The skill is knowing which one you are in and switching cleanly.

Shared Leadership Was the Beginning, Not the End

Pearce and Conger gave organizations a way to see what a century of leader-centric research had made invisible: the lateral web of influence that runs between teammates and often matters more than the line to the boss. That reframe explained why two teams with equally good managers can land in completely different places, and the meta-analyses confirmed that on complex, interdependent work the distributed web does more than the hierarchy. The claiming-and-granting research made the formation visible, and the network metrics made the distribution measurable.

What the model could not do on its own was tell you why a person steps up to lead, or which lever makes the network grow instead of stall. The Octalysis crosswalk supplies that, and with it the order of operations that turns the idea from a slogan into a design: build the CD1 purpose, the CD3 voice, and the CD5 safety first, preserve individual ownership underneath, and only then distribute the authority. Do it in that order and leadership distributes into a jazz quartet, each member taking the lead exactly when the work needs their voice. Do it backward, delete the managers and hope, and you get a traffic jam that calls itself flat. The model tells you leadership can be shared. The design discipline is refusing to confuse deleting the boss with building the conditions that let a team actually lead itself.

Frequently Asked Questions

What is shared leadership in simple terms?

Shared leadership is when leadership is distributed among the members of a team rather than held by a single appointed leader, with influence moving to whichever member has the relevant expertise for the task at hand. Pearce and Conger formalized the concept in 2003, defining it as a dynamic, interactive influence process among team members aimed at leading one another toward shared goals.

How is shared leadership different from vertical leadership?

Vertical leadership flows downhill from one formally appointed leader to followers, which is fast and clear and concentrates accountability. Shared leadership runs laterally, member to member, distributing influence across the team. They are complementary rather than opposed: the strongest evidence supports combining a vertical leader who sets direction with a dense lateral network that distributes execution.

What conditions does shared leadership need to work?

Carson, Tesluk, and Marrone identified three enabling conditions in 2007: shared purpose (a common internalized sense of what the team is for), social support (members actively encouraging each other’s contributions), and voice (genuine member input into how the team operates). Psychological safety underpins all three, because claiming leadership is an interpersonal risk members only take when the team feels safe.

Does shared leadership actually improve team performance?

Meta-analyses by Wang, Waldman, and Zhang (2014) and D’Innocenzo, Mathieu, and Kukenberger (2016) found that shared leadership predicts team effectiveness above and beyond vertical leadership, with the strongest effects on complex, highly interdependent tasks and in teams that have had time to mature. On simple, independent work, the coordination cost can outweigh the benefit.

When does shared leadership fail or backfire?

It fails when authority is distributed without the enabling conditions in place. Without shared purpose, freed influence becomes a status contest. Without psychological safety, it triggers accountability diffusion, the bystander effect where everyone assumes someone else will lead. It is also a poor fit for crises that need one fast, accountable decision, where vertical leadership is correct.

How do you measure shared leadership?

The standard method uses social network analysis. You ask every member who they turn to for leadership on which dimensions, then compute the network’s density (how many leadership ties exist) and centralization (how concentrated they are). Real shared leadership shows high density and low centralization. This catches the flat organization whose leadership is actually run by one informal alpha.

Is shared leadership the same as having no leader?

No. A leaderless team is usually a stuck team. Shared leadership means everyone might lead depending on the moment, not that nobody does, and it works best alongside a vertical leader whose job becomes growing the lateral network rather than supplying all the influence. The formal leader still matters; their role shifts from performing leadership to cultivating it.

How does shared leadership relate to the Octalysis Framework?

Shared leadership maps to a stack of Core Drives: it is a Core Drive 4 (Ownership) and Core Drive 5 (Relatedness) network property, gated by Core Drive 1 (shared purpose) and Core Drive 3 (voice), with leadership routed by Core Drive 2 (competence). Its failure mode is Core Drive 8 (Loss & Avoidance) accountability diffusion. The design rule that follows is to build those White Hat conditions first, then distribute authority on top of preserved individual ownership.

References

• Pearce, C. L., & Conger, J. A. (Eds.). (2003). Shared Leadership: Reframing the Hows and Whys of Leadership. Sage Publications.
• Pearce, C. L., & Sims, H. P. (2002). Vertical versus shared leadership as predictors of the effectiveness of change management teams. Group Dynamics: Theory, Research, and Practice, 6(2), 172–197.
• Pearce, C. L. (2004). The future of leadership: Combining vertical and shared leadership to transform knowledge work. Academy of Management Executive, 18(1), 47–57.
• Carson, J. B., Tesluk, P. E., & Marrone, J. A. (2007). Shared leadership in teams: An investigation of antecedent conditions and performance. Academy of Management Journal, 50(5), 1217–1234.
• Wang, D., Waldman, D. A., & Zhang, Z. (2014). A meta-analysis of shared leadership and team effectiveness. Journal of Applied Psychology, 99(2), 181–198.
• D’Innocenzo, L., Mathieu, J. E., & Kukenberger, M. R. (2016). A meta-analysis of different forms of shared leadership-team performance relations. Journal of Management, 42(7), 1964–1991.
• Nicolaides, V. C., LaPort, K. A., Chen, T. R., Tomassetti, A. J., Weis, E. J., Zaccaro, S. J., & Cortina, J. M. (2014). The shared leadership of teams: A meta-analysis of proximal, distal, and moderating relationships. The Leadership Quarterly, 25(5), 923–942.
• Mehra, A., Smith, B. R., Dixon, A. L., & Robertson, B. (2006). Distributed leadership in teams: The network of leadership perceptions and team performance. The Leadership Quarterly, 17(3), 232–245.
• Ensley, M. D., Hmieleski, K. M., & Pearce, C. L. (2006). The importance of vertical and shared leadership within new venture top management teams. The Leadership Quarterly, 17(3), 217–231.
• DeRue, D. S., & Ashford, S. J. (2010). Who will lead and who will follow? A social process of leadership identity construction in organizations. Academy of Management Review, 35(4), 627–647.
• Day, D. V., Gronn, P., & Salas, E. (2004). Leadership capacity in teams. The Leadership Quarterly, 15(6), 857–880.
• Gronn, P. (2002). Distributed leadership as a unit of analysis. The Leadership Quarterly, 13(4), 423–451.
• Gibb, C. A. (1954). Leadership. In G. Lindzey (Ed.), Handbook of Social Psychology (Vol. 2). Addison-Wesley.
• Bowers, D. G., & Seashore, S. E. (1966). Predicting organizational effectiveness with a four-factor theory of leadership. Administrative Science Quarterly, 11(2), 238–263.
• Hoch, J. E. (2013). Shared leadership and innovation: The role of vertical leadership and employee integrity. Journal of Business and Psychology, 28(2), 159–174.
• Friedrich, T. L., Vessey, W. B., Schuelke, M. J., Ruark, G. A., & Mumford, M. D. (2009). A framework for understanding collective leadership. The Leadership Quarterly, 20(6), 933–958.
• Edmondson, A. C. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350–383.
• Latané, B., Williams, K., & Harkins, S. (1979). Many hands make light the work: The causes and consequences of social loafing. Journal of Personality and Social Psychology, 37(6), 822–832.
• Leotti, L. A., & Delgado, M. R. (2011). The inherent reward of choice. Psychological Science, 22(10), 1310–1318.
• Lieberman, M. D. (2013). Social: Why Our Brains Are Wired to Connect. Crown.
• Chou, Y. (2015). Actionable Gamification: Beyond Points, Badges, and Leaderboards. Octalysis Media.

Related Reading

• Leader-Member Exchange — the vertical-tie counterpart to shared leadership’s lateral web.
• Employee Engagement — the outcome that psychological safety and voice feed into.
• Job Demands-Resources Model — how autonomy and support function as resources, not perks.
• The Octalysis Framework — the eight Core Drives behind the shared-leadership crosswalk.
• The Behavioral Framework Library — every psychological model in this series, mapped to behavioral design.

WOULD YOU LIKE YU-KAI CHOU TO WORK WITH YOUR ORGANIZATION?