Value Creation from Data - Why is this a BPM Problem?

Shazia Sadiq, Ida Asadi Someh, Tianwa Chen, Marta IndulskaProblems@BPM 2021 · OpenReview

Abstract

While data science and artificial intelligence promise a multi-trillion dollar economic revolution, many organizations struggle to translate raw data into tangible business value. This paper argues that the primary bottleneck is not technical, but socio-technical. By framing data value creation as a Business Process Management (BPM) challenge, organizations can bridge the gap between analytical insights and operational execution, weaving data directly into the fabric of organizational workflows.

The "Value Gap" in Data Initiatives

Despite massive investments in AI and machine learning, a significant "value gap" persists. Organizations frequently encounter three primary hurdles that prevent data from becoming a productive asset:

  1. Data Re-purposing: Data is often utilized for objectives far removed from its original design intent. This creates a "semantic distance" that challenges traditional data modeling and leads to misaligned insights.
  2. The "80% Tax": Data scientists continue to spend the vast majority of their time on manual data cleaning and preparation. This inefficient allocation of resources stems from a lack of standardized, repeatable processes for data curation.
  3. The Trust Gap: Complex, "black-box" analytical pipelines often lack transparency. This creates friction between technical teams and business stakeholders, who may be hesitant to act on results they do not fully understand or trust.

These issues are fundamentally organizational and process-oriented rather than purely algorithmic.

A BPM-Centric Research Agenda

To address these hurdles, this position paper proposes a shift in perspective, mapping established BPM concepts to modern data science challenges. The authors identify four specific opportunities where process-centric thinking can alleviate technical bottlenecks:

  • Process Orientation: Moving away from centralized, isolated analytics units toward cross-disciplinary teams embedded within business units. This ensures that data initiatives are aligned with actual business needs.
  • Agile Process Design: Developing frameworks to rapidly embed data insights into operational tasks, ensuring that analytics lead to immediate action rather than static reports.
  • Process Mining for Data Work: A novel application of process mining involves analyzing the workflows of data scientists themselves. By using event logs to understand how humans interact with data, organizations can discover more efficient strategies for data curation and identify where automation is most effective.
  • Reference Models: Moving away from ad-hoc, error-prone practices by developing standardized "blueprints" for data curation and AI development.

Key Insights: Data as a Process Component

A central thesis of this work is that data value is contextual. Data assets only generate value when they are integrated into decision-making points within a workflow. A dashboard that exists outside of a business process is significantly less valuable than an insight embedded at the moment a critical decision is made.

Furthermore, the authors highlight the Socio-Technical Barrier. The most significant risks to AI—including algorithmic bias and lack of explainability—often stem from inadequate handling of data complexity and a lack of transparency in the data pipeline. BPM provides the tools to manage this complexity by treating the data pipeline as a governed, visible process.

Finally, the paper addresses the Human-Machine Configuration. The "optimal" process is not always full automation. BPM research helps identify the right balance between augmenting human creativity and leveraging algorithmic speed, ensuring that AI serves to empower rather than replace the human element in complex workflows.

Conclusion and Future Directions

The field of BPM possesses significant "knowledge capital" that is currently underutilized in the data science domain. By framing value creation as a four-dimensional problem—encompassing organizational structure, agility, behavioral understanding, and standardization—the BPM community can lead the way in responsible and agile value creation.

While this paper serves as a call to action, future work must focus on developing specific tools and methodologies. This includes creating practical guidelines for human-AI interaction and testing regulatory frameworks to ensure they are adequate for the complexities of modern, data-driven business processes.

Summary updated

BibTeX
@inproceedings{sadiq2021value,
  author = {Shazia Sadiq and Ida Asadi Someh and Tianwa Chen and Marta Indulska},
  title = {{Value Creation from Data - Why is this a BPM Problem?}},
  booktitle = {Problems@BPM},
  year = {2021},
  url = {https://openreview.net/forum?id=s9fA3yQyMj},
}
Figure