Business Process Crowdsourcing: Beyond Micro-tasks to Enterprise Efficiency

TL;DR

Crowdsourcing is often viewed as a "low-end" solution for tagging photos or logo contests. This paper challenges that perception, proposing Business Process Crowdsourcing (BPC): a structured, managed alternative to traditional Business Process Outsourcing (BPO). By treating the "crowd" as an elastic cloud resource and applying formal orchestration, the authors demonstrate how enterprises can handle complex support and internal workflows with the same rigor—but lower cost—as professional outsourcing.

Problem & Motivation: The "Wasteful" Nature of Modern Crowdsourcing

Current crowdsourcing exists in three main flavors, all of which the authors argue are insufficient for the core business of a large enterprise:

1. The Contest Model: High waste; many work, only one is paid.
2. Task Marketplaces (e.g., MTurk): Limited to "brainless" repetitive tasks.
3. The Bid Model: Basically a digital version of traditional contracting, lacking elasticity.

The pain point is clear: Large enterprises refuse to outsource core logic because they lose control. Without a way to guarantee quality and integrate the crowd's output into existing ERP/CRM systems, crowdsourcing remains a peripheral toy rather than a strategic tool.

Methodology: The "Cloud of Crowd" Architecture

The core insight is to view human workers ($W$) not as temporary help, but as a computational node in a cloud infrastructure.

The Formal Framework

The authors define a general crowdsourcing system $C(T,W)$ where:

• $T$: A complex task split into manageable sub-tasks $t_i$.
• $K$: A set of constraints (skills, time, domain knowledge).
• $W$: A dynamic pool of workers whose expertise is tracked and updated.

(Note: This represents the orchestration layer connecting the task pool to the distributed worker network)

Unlike traditional BPO, which requires "hiring for peaks," the Cloud of Crowd allows for a fine-grained, pay-per-use model. The system is responsible for rescheduling in case of worker failure and using majority voting to ensure quality when workers are untrusted.

Experiments: Transforming Customer Support

The paper applies this to high-volume Customer Service. Using the "90-9-1" rule of online communities (90% lurkers, 9% intermittent, 1% heavy contributors), the authors calculate the deflection potential for a Telco with 10 million customers:

• Community Size: 2 million online users.
• Active Contributors: 220,000 potential "agents."
• Impact: Achieve a 20% deflection rate from traditional call centers to the community.

This isn't just "volunteering"; it's a managed process where the best contributors are algorithmically assigned to specific tickets ($R_i$), and their performance is recorded to build a "Dynamic Skill Profile."

(Note: The data illustrates how deflection rates scale with community engagement)

Critical Analysis & Conclusion

Takeaway

The true value of this work lies in the formalization of the crowd. By moving away from "open calls" and toward "orchestrated assignments," the authors provide a roadmap for enterprises to access global talent without sacrificing the KPIs (Key Performance Indicators) that run their business.

Limitations

• The Trust Gap: While "majority vote" works for simple tasks, it is difficult to apply to highly subjective or creative business processes.
• Incentive Alignment: The paper assumes that rewards and gamification can consistently motivate a workforce that isn't under a formal employment contract.

Future Outlook

As we move into an era of AI-Human collaboration, the BPC model is more relevant than ever. Future systems will likely see the "Cloud of Crowd" replaced by a "Cloud of Agents," where human experts act as the final quality gate in an orchestrated loop of automated sub-tasks.