Introduction

In the first part of this article, we introduced data products and looked at some use cases of data products in financial services. But execution of a successful data product strategy hinges on defining the right data product operating model.

A robust operating model provides LOBs and IT divisions a framework or common language to collaborate effectively for building and managing data products. It ensures that data product delivers the intended benefits envisaged by business.

In this article we will share our learnings about how organizations can define, build and manage data products. This is based on our experience from rolling out an enterprise-wide operating model for a major banking client.

First, let’s discuss what data product operating model is

It is a set of guidelines that define how data products are developed, managed and maintained within the enterprise. It provides a structured approach across people, process, tools and governance parameters to ensure data products align with business objectives, meet quality standards, and deliver value to consumers. In the image below, we’ve highlighted key components of data product operating model.

While details on above components could be organization specific, we’ve summarized some key principles below that can guide the blueprint for defining these components.

1.      Data should be segmented into domains across the organization – Data which is mostly in centralized monolithic warehouses must be broken down and organized into business domains such as mortgages, credit cards, credit risk, customer 360 insights data products etc. LOBs must take full ownership and accountability for creation and maintenance of data products and ensure that they align closely with business objectives and user needs. Some key points for business domains/LOBs to consider are –

• They are accountable for the reliability, availability, and accuracy of the data as the domain experts of the data.
• They control access to data product ensuring appropriate usage and consumption across the organization.
• They should drive the prioritization of data products and fund development and maintenance of data products, ensuring that investments are aligned with business priorities, use cases and ROI measures.
• They must implement and adhere to data governance practices, including privacy compliance, security measures, and data retention policies for their data products.

2.      Access to data products must be democratized – Like an Amazon or Alibaba marketplace, imagine a data product marketplace where consumers from anywhere in the organization can browse data products, view product specification, see/write reviews and request access from the data product manager. A marketplace model gives data consumers readily available, high quality, trusted data assets democratizing data access to anyone in the enterprise and significantly accelerating time to insights. Monetization, wherein the data product owner charges data consumers based on no. of users or transaction volumes etc. can also be explored to self- fund opex. costs for data products. Access management to data products must be controlled by LOBs and data product managers.

3.      Central IT should provide the infrastructure for business to develop data products – IT should provide the necessary data infrastructure, tools, and platforms for creating and managing data products. They must offer expertise in data architecture, cloud-based data solutions, and data security. While collaborating closely with business units, this specialised IT team focuses on the technical aspects of data product delivery, leaving ownership and strategic decisions of data products to the business.

4.      A data COE must be established – Establishing a dedicated data COE to standardize data product development and promote best practices is essential for the organization. Establishing standards and best practices includes defining data architectural pattern, templates and tooling for data product design and development. To ensure data governance and consistency, the Data COE defines protocols for data provenance, audit trails, and quality metrics. Furthermore, it architecturally integrates necessary technologies for each consumption archetype, promoting reuse in all data products.

5.      Dedicated resourcing and funding must be allocated – Every data product must be supported by a dedicated, funded team, led by a product manager and consisting of data engineers, architects, modelers, and platform engineers, to drive development and innovation.

Data product managers especially should sit within LOBs and their primary role is to support firms’ and their division’s strategic objectives. Data product manager should have broad range of skillsets encompassing domain, product management, data analytics and platforms. Like any IT product manager, a data product manager defines business needs, collaborates with data engineers, architects and data scientists to translate business needs into data product solutions, monitor the ROI of those data products and enhance it iteratively. Additionally, they promote data literacy in LOBs and articulate the value unlocked through usage of data products.

The principles above could be challenging to implement particularly in organizations that have always adopted traditional methodologies for their data. In the section below, we have highlighted some key learnings for CDOs and CIOs who are building data products in their firms.

What can derail data product initiatives in firms and how to mitigate them

Conclusion

Successful data product implementations require a robust data product operating model. It could really be the difference maker helping firms navigate through organizational complexities and enable them to derive business value from data products.