Introduction

In today’s enterprises, data serves as an essential asset for managing knowledge and informing decision-making. Yet, the process of transforming data into actionable insights can become fragmented, hindered by ad hoc collaboration and unclear ownership. The Insights Supply Chain provides a structured, knowledge-driven approach to addressing these challenges by integrating data workflows across specialized, yet sometimes overlapping, roles. The Insights Supply Chain serves as a knowledge management framework that systematically integrates data expertise into decision-making.

Since I created the concept of the Insights Supply Chain, empirical research has appeared to directly support its core principles of structured collaboration. In “How do Data Science Workers Collaborate? Roles, Workflows, and Tools”, Zhang, Muller, and Wang (2020) provide valuable insights into how data science teams collaborate to generate business value. The researchers staked claims to publishing several firsts in the study of data science work including a comprehensive exploration of collaboration in data science teams, and a stage-based analysis of activities at scale and of collaborative tool usage. When analyzed through the lens of the Insights Supply Chain, these findings further validate the importance of structuring data science collaboration within a broader knowledge management framework.

The Role of the Insights Supply Chain in Integrating Data Science Workflows

The Insights Supply Chain is a structured workflow in which data moves through distinct stages – from collection and analysis to insight generation and decision-making. Unlike practices that rely on isolated hand-offs between teams, this framework focuses on integrating workflows to ensure a seamless transition between specialized roles that often overlap in practice:

• Data Engineers build and maintain infrastructure, optimizing computing resources and ensuring data accessibility through efficient data flows.
• Data Analysts extract, clean, and structure data to reveal patterns and inform modeling.
• Insights Analysts translate and synthesize analytical findings into compelling business recommendations that support decision-making.

Data scientists can function in any or all three of these roles depending on a company’s organizational structure.

Zhang et al. (2020) identified six key stages of data science work: “create a measurement plan; access, combine, and/or clean data; select, extract, and/or engineer features; train and apply model(s); evaluate outcomes; communicate with clients and/or stakeholders”. Embedding these stages within the Insights Supply Chain involves:

• The identification of a problem to solve (from anywhere in the organization) initiates the creation of a measurement plan.
• A measurement plan defines and launches a Data Engineering effort to identify and collect data in as close to model-ready form as possible.
• Based on available data, the Data Analyst conducts modeling and analysis which includes any combination or sequence of feature engineering, model optimization, and evaluation.
• The Insights Analyst uses the model results to motivate actions and decisions aligned with the problem.

At each stage, technical and business expertise must be aligned to keep the Insights Supply Chain focused on the right problem and the right decisions. The degree of collaboration required depends on how an organization structures its data teams. Effective knowledge management relies on these structural choices, determining how data expertise is distributed and leveraged across teams. The Data Org Matrix (DOM) provides a framework for making these decisions, balancing generalization vs. specialization and centralization vs. decentralization.

In the DOM, collaboration occurs in data organizations with data engineering, data analytics, and data science as well-defined roles. The locus of collaboration depends on the level of centralization of these roles. Because Zhang et al. (2020) study practices within a single company (IBM), the organizational structure is taken as a given and not considered as a salient feature driving modes of collaboration. The intensity and frequency of links in the study’s network graph of reported collaborative relationships must be influenced by the organizational distance of collaborators whether through degree of specialization or centralization. For example, the study observes “…strong intra-role collaborations among Engineers, Researchers, and Communicators. By contrast, Managers/Executives and Domain Experts appear to collaborate less with other members of their own roles.” The Insights Supply Chain identifies underlying organizational structures that help to shape those interactions. The Insights Supply Chain makes explicit the organizational dependencies for achieving data work.

Organizational dependencies also determine how knowledge is documented, shared, and applied. For highly specialized organizations, knowledge management tends to be more structured within domain-specific teams, whereas in generalist organizations, knowledge must be more widely accessible across functions to enable cross-disciplinary collaboration. In an organization of generalists, each person or team actively seeks the help and knowledge of others because responsibilities in a project cannot be limited by isolated knowledge bases. In an organization of specialists, knowledge management also looks more specialized as project responsibilities more neatly fit within the bounds of particular expertise. The knowledge management process deepens the domain expertise of individuals and teams. Collaboration occurs from specialized needs and knowledge is groomed by domain.

Decentralization vs. Centralization in Data Science Collaboration

The specialization of data roles within an organization naturally raises a structural question – should these roles be managed within a centralized team or distributed across decentralized units? An organization’s degree of specialization also influences whether a centralized or decentralized structure is more effective.

Let’s examine more closely the implications of the x-axis of the DOM. The DOM enables a company to determine how and where data science should be embedded in the business. A key consideration in structuring data science teams is whether to centralize or decentralize data expertise. This choice has implications for collaboration efficiency, decision-making speed, and alignment with business strategy.

Centralized data science teams tend to develop standardized processes, maintain consistency in methodologies, and provide stronger governance over data quality and compliance. Data professionals in centralized teams tend to have clear career trajectories given their proximity to peer experts. However, centralization can also create bottlenecks if business teams and other stakeholders must wait for approval or prioritization from a centralized unit. By contrast, decentralized teams allow for greater agility from the perspective of business teams and other stakeholders. Insights can be quickly tailored to specific business needs. However, decentralization can lead to duplication of effort and inconsistency in methodologies across different teams. Moreover, decentralized teams require coordination at the team level, which is more resource-intensive than managing collaboration within a centralized structure.

The Insights Supply Chain provides transparency to the organizational choices underlying data science collaboration, ensuring that whether an organization chooses centralization or decentralization, there is a clear knowledge management framework in place. Zhang et al. (2020) implicitly assume a single organizational model, but different companies may require different structures based on their industry, scale, and business strategy. By providing visibility into these trade-offs, the Insights Supply Chain drives informed decisions about data strategy, balancing speed and governance based on business needs.

Integrating Data Science into Organizational Structure: Is Data a Core Competence?

Another critical factor in deciding how to structure data science collaboration is determining whether data is a core competence of the organization. The DOM provides a framework for thinking through the choices and implications involved.

If data is a core competence, then the skillsets of data professionals need to be specialized to focus on a productized view of data flows. If data is primarily a support function, then generalized skills may be sufficient where data professionals are allocated according to the needs of specific business units. Either way, modes of collaboration will translate directly from decisions on core competency to the collaboration modes of the DOM.

The Importance of Knowledge Brokerage in Data Science Collaboration

Zhang et al. (2020) identifies the importance of intermediaries who facilitate communication between technical and non-technical collaborators because in certain cases “data science workers do not necessarily know how to translate this business question into a data science question.” This role, which they describe as a “brokering activity,” relies on a “special group of organizers who understand both data science and the context”. As a result, they “serve as translators to turn business questions into data science questions.”

In the Insights Supply Chain, this role fits within the purview of the Insights Analyst, whose responsibilities extend beyond data preparation and analysis to actively embedding insights into business processes and ensuring alignment with enterprise strategy. The Insights Analyst ensures data-driven insights are effectively transformed into enterprise decisions. Without an Insights Analyst role or function, data science teams may generate sophisticated models that never influence enterprise decisions, while business leaders struggle to translate strategic goals into measurable data initiatives. This disconnect leads to misaligned priorities, wasted analytical effort, and insights that fail to drive meaningful action.

Early in my career as a data professional, I played this brokering role while carrying the titles “Decision Engineer” and then “Solutions Architect”. I bridged the “Analytics Engineers” (today they would be called Data Scientists) who needed to create business-relevant models and marketing customers who needed to access actionable business insights from our pricing models.

This brokering activity is a reminder that the Insights Supply Chain does not start with data for data’s sake. The process starts with a problem to solve where data and related flows are filtered and modeled according to their relevancy to the problem at hand. While the Insights Analyst role sits at the end of the Supply Chain, it also heavily influences upstream work.

Conclusion: Data Science Collaboration In the Insights Supply Chain

Organizations looking to effectively integrate data science into enterprise decision-making must define a structured, knowledge-driven approach to managing data workflows. The Insights Supply Chain provides a framework for achieving this integration by:

1. Structuring data science workflows as a knowledge management process using the Data Org Matrix (DOM)
   • Enterprises should embed data science workflows into a broader system with explicit processes for managing knowledge. These processes must align with choices across specialization vs generalization and centralization vs decentralization.
2. Aligning organizational structure to data’s role in the business using the Data Org Matrix (DOM)
   • Companies must determine whether data is a core competence or a support function. This decision drives choices in the DOM for the structure of data teams.
3. Ensuring collaboration is a product of intentional design
   • Collaboration in data science relies on designing processes that ensure insights flow efficiently between technical and business stakeholders.
   • Whether an organization centralizes or decentralizes its data science function, the Insights Supply Chain provides a framework for effectively managing this collaboration.
4. Recognizing and formalizing the role of knowledge brokers
   • The Insights Analyst (or equivalent role) should be formally recognized as a critical function in translating data insights into business decisions.
   • Organizations should invest in hiring, training, and supporting these professionals, ensuring that they have both technical fluency and business acumen.

The Insights Supply Chain transforms data science into a structured knowledge management process that drives enterprise decision-making. Organizations that embrace this framework will not only improve collaboration but also position data science as a fundamental pillar of business strategy.

The question then becomes: how do enterprises take the first step? To begin implementing the Insights Supply Chain in your organization, start by evaluating your current data workflows through the lens of the Data Org Matrix. Identify where structural inefficiencies exist, and assess whether your current degree of specialization and centralization align with your enterprise’s business strategies and goals.

For more information on how Ahan Analytics can help you accelerate this transition, contact us today.

References

Zhang, A. X., Muller, M., & Wang, D. (2020). How Do Data Science Workers Collaborate? Roles, Workflows, and Tools. Proceedings of the ACM on Human-Computer Interaction, 4(CSCW1), Article 22. https://doi.org/10.1145/3392826

---