Data subsets and data products Part 2: Economics and journey maps

Vehicle subassemblies

In Part 1 of “Building Blocks of Modern Data Management,” I explored two important concepts of modern data management: data subsets and data products (Figure 1).

  • Data subsets are the packaging and pre-wiring of data and its media (e.g., enriched metadata, data access methods, data governance policies and procedures, data access security protocols, data quality scores, privacy rules and regulations, usage patterns) in a single “package” to simplify data discovery, access and exploration to optimize the efficiency and productivity of data workers (e.g. data engineers, data scientists, business analysts).
  • Data products are a packaging of AI/ML analysis and customer, product, service and/or operational analytical insights (predicted behavioral and performance propensities) as an application to help end users (non-data workers) achieve specific business or operational results.
Slide 1-1

Figure 1: Data Subset vs. Data Product

As a concrete example, automotive sub-assemblies have been instrumental in product consistency, reliability, quality, profitability and scale in the automotive industry by increasing production efficiency, reducing the final production lead time, reducing manufacturing risks, reducing assembly failures. , improving quality and reliability and reducing costs associated with labor, manufacturing, procurement, inventory, logistics, maintenance and support (Figure 2).


Figure 2: Automotive sub-assemblies

In Part 2 of this series, I want to dive deeper into the economics of data subsets and data products. I also want to explore the important role of these two data management “products” in accelerating an organization’s data management journey from business need to business outcome.

The reason I separate data subsets from data products is that the same data subsets can be reused in the development of multiple “data products”. For example, a “point-of-sale (POS) data subset” – point-of-sale data aggregated with its metadata, access methods, data governance policies, access security protocols, quality scores data, etc Data products including:

  • Merchandising data product used to make in-store merchandising and pricing recommendations based on store and customer buying propensities.
  • Customer Loyalty Data Product used to report and make recommendations to resolve potential customer churn issues.
  • Customer acquisition data product used to make customer targeting and promotion recommendations based on “like” customer profile behaviors and purchase propensities.
  • Customer Up/Cross-Sell Data Product used to make product recommendations based on individual customer purchases and product behaviors and preferences.
  • Inventory Management Data Product used to make stocking and markdown recommendations aimed at reducing stock-outs and optimizing markdown management.
  • New Product Introduction Data Product used to make merchandising, promotion and pricing recommendations for new products that increase the likelihood of successful new product introductions.
Slide 3-1

Figure 3: The economics of data subsets

If these data subsets are designed for sharing, reuse, and continual refinement, reuse of these data subsets can occur at near-zero marginal cost, making them a very powerful digital economic asset. Additionally, any improvements in the quality, accuracy, completeness, latency, granularity, and enrichment of data subsets trickle down to every data product that reuses that same subset. of data, feeding Schmarzo’s economic digital asset valuation theorem (Figure 4).


Figure 4: Schmarzo’s Economic Digital Asset Valuation Theorem

In the blog “Fallacy of Becoming Data-driven – Part 1: Becoming Value-obsessed”, I introduced the Data Management Candy Land Journey. The objective of the Data Management Candy Land Journey was to highlight key data management outcomes and execution dangers, unlocking the economic value of an organization’s data by accelerating the data management journey of need business to the business outcome (Figure 5).


Figure 5: Candy Land Journey Data Management

I got wonderful feedback on how organizations were using this journey map to guide their data management efforts (but wanted a version that could be shared with company management). Therefore, I created a simpler (less fun) version that could be used with business executives to describe key data management results needed to accelerate an organization’s data journey business need at Business result (Figure 6).


Figure 6: Data Management Value Journey Map –Version 2.0

Some key takeaways from version 2.0 of the data management journey map:

  • Guide the transition of organizations data projects reusable, refinable data products. The updated data product development canvas and the new data assembly development canvas will be covered in my next blog in this series.
  • Data subsets and data products are economic assets that enable organizations to accelerate and scale their data-driven, value-creating data management journey from business need to business outcome.
  • The ability to share, reuse and constantly refine data subsets and data products distinguish these products as essential enablers to unleash the business or economic value modern organization data.
  • The data management journey is a continuous cycle where business results feed into the next iteration of the journey. And if properly instrumented with Data Observability, we can create a data management journey that leverages AI/ML to continuously learn and adapt with minimal (semi-autonomous) human intervention. See my blog “Data Observability and Its Importance in Setting Intent” for more details on how to leverage data observability to create continuous learning and adaptation processes.
  • The image behind the Data Management Journey Map version 2.0 is the Data & Analytics Business Model Maturity Index. Why? Because ultimately, the goal of any modern organization is to become more efficient at leveraging data and analytics to power its business and operating models. Yeah, that’s what it’s all about.

Data subsets and data products as packaged, shareable, reusable, and continuously refinable assets that enable organizations to become more efficient in scaling the data and analytics economy across the organization.

In my next blog in this series, I will introduce the updated data product development canvas. Several people have offered to review this next iteration, so it should be pretty good.


About Author

Comments are closed.