Break down silos. Maximize connectivity. Get everyone in the room. This is the default advice for innovation. And sometimes it works well. When Ford and GE Healthcare needed to produce 50,000 ventilators in 100 days during the Covid-19 pandemic, intensive cross-functional collaboration was exactly right. Every decision was interdependent, every insight shared, every problem solved collectively. However, apply that same approach to a different context, and you will not only fail to innovate but also systematically prevent innovation from happening. When NASA needed to predict dangerous solar particle events, the breakthrough didn’t come from traditional collaboration. It came from an open innovation challenge where Bruce Cragin, a semi-retired engineer working alone in rural New Hampshire, explored a solution path that NASA’s teams would never have considered. Same imperative. Opposite approaches. Both successful. This is not a paradox. When we analyzed 294 empirical studies on collective innovation, we found that connecting people to collaborate doesn’t have a consistent effect on innovation outcomes. Sometimes it’s strongly positive. Sometimes it’s negative. Sometimes it has no effect at all. In some cases, the relationship isn’t even linear; moderate connectivity might help in a situation where high connectivity would hurt. The reason for these varied effects lies in something most innovation advice overlooks: the structure of the collective itself. Innovation emerges under different conditions depending on how collectives are structured, how members interact, and how aligned their goals are. Different Types of Collectives Managing collective innovation begins with a question most leaders skip: What kind of collective are you actually working with? Organizations often treat all collaborative innovation as if it requires the same approach. Yet our review reveals that collectives differ in two fundamental ways that determine how they should be managed: Search dependence: How interdependent is the process of finding solutions? Do people need to build on each other’s work as they explore possibilities, or can they search for solutions independently? Goal alignment: How much are participants working toward the same overarching objective versus pursuing their own priorities? Based on these two dimensions, three distinct types of innovation collectives emerge: convergence-based (interdependent search + shared goals), divergence-based (interdependent search + unshared goals), and attention-based (independent search + shared goals). The remaining case—independent search with unshared goals—isn’t really a collective at all; it’s just actors operating in parallel. Convergence-Based Collectives: When innovation problems require tight integration of expertise, convergence-based collectives bring the right people into the same room, literally or virtually, to solve problems together. Members build on each other’s insights in real time, providing feedback, challenging assumptions, and co-creating solutions. This is the model for cross-functional product development teams or interdisciplinary R&D efforts. For example, when Alphabet’s DeepMind needed to predict protein structures—a problem that had stumped scientists for nearly 50 years—they assembled machine-learning engineers, biologists, and physicists to build AlphaFold. The breakthrough required continuous synthesis of insights across disciplines, something no individual or independent group could have achieved. Divergence-Based Collectives: Sometimes the best innovation comes from multiple entities pursuing their own paths within a shared ecosystem. In divergence-based collectives, participants have different goals but benefit from information flow. They might compete, collaborate selectively, or simply learn from each other’s experiments. A case in point is Johnson & Johnson’s JLABS network. Across 13 sites, more than 650 biotech startups pursue their own ideas with “no strings attached,” and all gain from shared infrastructure, investor connections, and the broader ecosystem’s knowledge. Each company follows its own agenda, yet the whole network becomes more innovative than the sum of its parts. Attention-Based Collectives: For well-defined problems where multiple solution paths are possible, attention-based collectives harness independent work at scale. Participants share a common goal but work in parallel, exploring different approaches without coordinating their thinking. The best solutions emerge through selection rather than collaboration. When Meta needed to tackle deepfakes in 2020, they launched a $1 million challenge that attracted over 35,000 submissions. Each participant worked independently, testing different detection models. The winning approach reached 82.56% accuracy, and Meta gained access to thousands of alternative solutions they could never have generated internally. Promoting Innovation in Divergence-Based Collectives Convergence-based collectives unlock innovation that no single contributor could achieve alone, but they demand more from managers: more resources, more coordination, and more deliberate design. Success depends on two critical processes: accessing diverse knowledge and integrating it into coherent solutions. It starts with the composition of the collective. Most research suggests that assembling collectives with a wide range of disciplinary backgrounds and cognitive styles leads to stronger creative outcomes. But it is important to prioritize diversity in professional backgrounds and expertise rather than relying only on demographic differences. Take how the IDEO structures its teams. The global design and innovation firm composes teams mixing business strategists, engineers, anthropologists, and designers. This cognitive diversity allows teams to approach problems from multiple angles, fostering the cross-pollination that has helped IDEO create breakthrough products for clients from Apple to the Mayo Clinic. The value of that diversity, however, depends on the ability to bridge it. Research shows that cognitive diversity enhances creativity when collaboration dynamics encourage the constructive use of differing perspectives. To facilitate this integration, managers must build an environment of trust, psychological safety, and open communication. To create such an environment, leadership plays a key role. For example, several studies show that transformational leadership and rotating leadership—shifting decision-making authority among team members based on task needs—significantly promoted team innovation by using distributed knowledge more effectively. This was primarily due to strengthened trust, participation, and collaboration across the team. Another useful approach is incorporating collective performance-based incentives, such as profit-sharing or team-based rewards, to reinforce these behaviors. For example, a study of 22,000 European firms showed such incentives enhanced organizational innovation by strengthening psychological ownership, information sharing, and teamwork. Haier organizes itself into thousands of “microenterprises,” where any employee or team can launch a venture and hold shares in it—creating value and directly sharing in the profits. This model has yielded 183 startups within the company, accelerating new product rollouts beyond what traditional structures typically achieve. Promoting Innovation in Divergence-Based Collectives Divergence-based collectives thrive on loosely coordinated exploration. Semi-autonomous units—often across firms or departments—pursue their own innovation paths while embedded in a broader network. This model works when the path forward is unclear and the challenge benefits from parallel experimentation across different domains, such as emerging technologies or industry-wide transformations. The managerial challenge is threefold: building the right network, avoiding information overload, and developing the capacity to absorb what flows in. This is often a fragile balancing act. More partners mean more perspectives until they mean chaos. Left unchecked, a sprawling network can bury teams in redundant information, dilute focus, and slow decision cycles. Leaders must therefore orchestrate relationships that deliver novel information, shield the collective from overload, and build internal capabilities to absorb what flows in. The richest innovation contributions come from partners who see the world differently. Curating strategic partnerships with entities that bring novel insights rather than redundant knowledge is critical. For example, a study of Japanese supply chains found that firms with strong ties to geographically distant suppliers outperformed those embedded in local-only networks. These distant connections introduce fresh thinking. In contrast, an analysis of over 2 million inventors across 337 U.S. regions found that firms locked into dense local clusters were less exposed to novel information and experienced declining innovation productivity. But this has a limit. Distributing innovation efforts too broadly can also pose risks. A study analyzing more than 500,000 patents across 1,127 firms found that while geographically dispersed R&D teams accessed more diverse knowledge, they also faced steep integration costs, often reducing the quality of innovation output. Consider BMW’s intentionally selective approach in its Startup Garage, where the company screens over a thousand startups each year to fast-track only a few into its network. The approach has yielded success: it has completed projects with 220 companies across 26 countries, 30 of which became long-term suppliers. One way to overcome this burden of breadth is by developing absorptive capacity—i.e., systems and skills to process external ideas. Research consistently shows that companies with greater absorptive capacity benefit more from external collaboration. For instance, Novartis’s Biome hubs rapidly onboard external partners, pair them with in-house experts, and give them access to sandboxes of curated clinical-trial data. This strengthens Novartis’s ability to digest its vast incoming data streams, enabling faster, more informed innovation. Promoting Innovation in Attention-Based Collectives To unlock the full potential of attention-based collectives, managers must design environments that attract diverse contributors while preserving the independence that makes the model work. Outsiders are critical in attention-based innovation. In one study of 166 scientific challenges involving over 12,000 participants, contributors from outside the core domain were more likely to produce breakthrough solutions. However, novelty alone isn’t enough. Another study of 230 science contest submissions found that outsiders generated better results only when they first gained basic familiarity with the problem domain. The takeaway: invite people from unconventional backgrounds but equip them with enough context to orient their thinking. To preserve creative independence, leaders must avoid over-structuring the process. Research shows that exposing contributors to prior ideas too early can backfire, as it introduces cognitive anchors that narrow the solution space. In a series of experimental studies, increased peer visibility reduced originality, especially when individuals conformed to early submissions. The best outcomes often come when contributors work without the influence of others’ thinking. Incentives also matter, but not in the way many assume. A study of 646 solvers on a major crowdsourcing platform found that both intrinsic motivation (such as the joy of solving) and extrinsic motivation (financial rewards, recognition) played complementary roles in driving the quality of innovation solutions. The most effective designs often tap into both. When it comes to extrinsic motivation, the traditional winner-takes-all contests can discourage participation and diminish quality. A more effective model, according to a study of 261 innovation contests, is to offer multiple smaller prizes and allow repeat winners. This approach increased engagement, diversified contributions, and led to higher-quality solutions over time. Google’s Vulnerability Reward Program exemplifies this approach: in 2024, it paid $11.8 million to 660 researchers, with many earning multiple payouts for different discoveries. Feedback is also a powerful lever when used well. In two longitudinal experiments and a survey of over 1,500 innovation managers, constructive feedback that focused on actionable improvement significantly increased early-stage participation and led to stronger final outcomes. Encouraging iterative refinement turns isolated ideas into polished solutions. Even observing the feedback given to others helps. A study of over 55,000 participants on Threadless, a platform for t-shirt design contests and an e-commerce site, found that simply seeing how others were evaluated helped contributors calibrate their own responses. Over time, this feedback visibility improved learning, leading to improved solutions. Designing the Right Collective for Innovation Our research reveals something most innovation advice overlooks: there is no universally effective strategy for collective innovation. The managerial practices that drive breakthroughs in one collective type can systematically prevent them in another. Consider information sharing. When problems require integrated expertise, making everyone’s thinking visible is essential—real-time synthesis is how breakthroughs emerge. But when problems need independent exploration, that same visibility anchors thinking and narrows the solution space. Or take coordination. Convergence-based collectives require tight alignment because the work itself is interdependent. Attention-based collectives, however, thrive on autonomy—the diversity of solutions comes from contributors exploring different paths without influencing each other. The Ford-GE Healthcare ventilator effort and the NASA solar particle prediction both succeeded through collective innovation. Yet, they required opposite management approaches because they had different collective structures. . . . Most organizations default to convergence-based approaches because they feel productive—putting smart people together, maximizing information flow, driving alignment. Sometimes that instinct is exactly right. But when the problem actually requires independence or loose coordination, that same instinct becomes the constraint. The best innovation leaders recognize that “breaking down silos” is sometimes the answer and sometimes the problem. That distinction—between managing collaboration generically and architecting collectives strategically—separates innovation theater from innovation capability.
