download pdf

Not all crowdsourcing challenges are created equal
New information technologies have allowed companies to tap into the creative potential, distributed work patterns, and expansive knowledge of huge online crowds. In various business fields, crowds can solve certain problems faster, better, and cheaper than companies are able to do in house.  Today, according to a trend report published by the platform provider eYeka in 2015, 84 % of the world’s top companies – including SAP, Dell, Google, General Electric, Fiat, LEGO, and Procter & Gamble – have started to build their own crowdsourcing platforms.  The crowd-sourced tasks, however, are highly diverse, as are crowdsourcing platforms. For instance, the Fiat Mio platform, where contributors collaborated to develop a new concept car, is completely different from the GE Ecomagination Challenge, where contributors compete against each other. In the case of Fiat Mio, contributions were small and reflected by sharing, commenting, editing, or integrating ideas for further developing the car in a collaborative fashion. In contrast, GE’s Ecomagination Challenge does not require substantial collaboration among contributors. It facilitates an innovation contest in which each contribution reflects an independent and exhaustive solution to a specific crowdsourced task. Of course, the different nature of the tasks demands different governance mechanisms. While collaboration is an important issue for Fiat Mio, mechanisms that permit the control and evaluation of a high number of alternative contributions will be a key challenge for GE.

Different types of crowdsourcing platforms
Crowdsourcing platforms fall into four categories, distinguished by the diversity and aggregation of their contributions (see Figure 1). The main goal of microtasking crowdsourcing platforms is the scalable and time-efficient batch processing of highly repetitive tasks, e.g., categorizing data or writing and translating small chunks of text. Crowdsourcing platforms for information pooling aggregate contributions such as votes, opinions, assessments, and forecasts through approaches such as averaging, summation, or visualization. Broadcast search platforms collect contributions to solve a task to gain alternative insights and solutions from people outside the organization. They are particularly suited for solving challenging technical, analytical, scientific, or creative problems. Frequently, broadcast search is applied to running different kinds of innovation, design, or data science contests. Finally, open collaboration platforms invite contributors to team up to jointly solve a complex problem where the solution requires the integration of distributed knowledge and the skills of many contributors. The individual contributions are aggregated such that one or more solutions to the underlying problem can emerge. In practice, however, pure forms of these archetypes are rare. Frequently, crowdsourcing platforms combine several traits.

How to manage the different types successfully
The management of these different types needs to reflect their varied goals and nature along several[CK1]  dimensions. Governance involves structuring roles and responsibilities, formal and informal rules, standards and regulations, outcome control measures, communication processes, or details of task allocation to achieve a crowdsourcer’s goal. In a research project, we identified six distinct governance domains that encompass 21 distinct governance mechanisms for crowdsourcing. We investigated a total of 19 platforms and for each platform type, we studied at least four typical platforms. The purpose of our study was to identify effective governance mechanisms for each type of platform. Figure 2 summarizes which types of governance mechanisms are effective for the different types.

Effective governance of microtasking platforms
Organizations that host a microtasking platform should consider governance mechanisms that are primarily geared towards assuring an adequate quality of contributions. In order to ensure the repeated and parallelized execution of tasks, modularization is key; to receive high quality contributions, crowdsourcers should communicate contribution requirements. Such definitions provide contributors with a clear set of instructions to help them better understand the tasks and document the results of their work.  For example, Clickworker provides templates for defining the characteristics of desired results.  
The large quantity of contributions is also a challenge for monitoring quality. Whenever possible, management should opt for automated quality control. Our results demonstrate that relatively simple measures that are easily adapted by many crowdsourcers prove to be effective. Financial payments are the primary incentive for contributors. Typically, each contributor who completes a valid solution that satisfies the task requirements is paid. We recommend that hosts of microtasking platforms also install a reputation system with which contributors can signal their skills, expertise, or participation level. These systems effectively complement financial incentives and increase the effort exerted by contributors. Further, authentication for verifying the identity of contributors is recommended as it helps prevent misconduct. It removes the anonymity of contributors and incentivizes more accurate task performance.

Effective governance of information pooling platforms
Organizations intending to establish an information pooling platform should implement a governance structure that focuses on helping contributors submit high quality information. They should define contribution requirements and offer tutorials. For instance, BahnScout has clear guidelines: Contributors are expected to include a picture of the issue, a textual description, the precise location, select a predefined category, and mention potential hazards. Typically, contributors voluntarily participate in this type of crowdsourcing and therefore most contributors are personally interested in the task or project. To get a realistic picture and avoid bias, organizations should focus on integrating diverse and independent contributions, e. g. by demographic-based task allocation. For this type of crowd work, non-financial incentive mechanisms such as reputation systems are most effective. Rankings or experience levels are good tools to motivate contributors because they enable contributors to signal their standing within a platform’s community. Similarly, socialization enables contributors to communicate and interact with peers and is often appreciated.

Effective governance of broadcast search platforms
Completely open approaches to broadcast search tend to create a lot of “noise”, resulting in many low-quality contributions. In order to receive a manageable number of contributions without substantially reducing the chances of getting high quality, organizations should consider focusing their broadcast search on groups of contributors with proven abilities.  For this type of platform, contribution requirements again play a crucial role and should be defined carefully. They should ensure that results can be implemented in practice. For broadcast search, financial incentives are particularly important. Usually the best contribution receives a significant prize while unsuccessful participants come away empty-handed. For example, the jovoto platform recognized that competing for such prizes is perceived as risky by many contributors. To ensure broad participation, jovoto usually offers multiple prizes such as rewards for runner-up contributions or progress prizes for best contribution at the halfway point of the contest. In some cases, payments for participating can also be considered. This is common when a group of contributors with specific skills are included within the broadcast search, e.g., design professionals, or for invitation-only projects with a limited number of participants.