Scaling Data Value Creation: Five Principles To Ensure Data Marketplace Success

Chief data officers (CDOs) in financial institutions are working hard to remove a series of frictions that hamper the creation of value from data. One of the biggest is ensuring that data users across the institution can quickly discover the right data, then easily access it to generate insights and create innovative products and services.

The fundamental challenge is that data often remains in business and functional silos across the enterprise. Potential users need not only to know that the data exists but also to understand its quality and the purposes for which it can legitimately be used.

This is where data marketplaces come in. Modern data marketplaces introduce AI-assisted self-service, enabling users to discover and access data assets that are supported by quality assurance, metadata, appropriate permissions, and controlled, automated workflows.

However, it is easy for CDOs to go down the wrong path when building data marketplaces, especially if the challenge is conceptualized as simply finding the best technology ‘solution’. Building a data marketplace is a multi-faceted challenge. It includes transforming underlying data access and publishing processes as well as the enterprise data culture — in particular, practices around data sharing and building trust.

Furthermore, institutions have different starting points in terms of the maturity of their data environments and different priorities for their goals and use cases. In our experience, building the right data marketplace means leveraging five principles from the start:

1. Set out institution-specific goals tied to metrics. First, define the key organizational goals clearly, and partner with business domains to define the business priorities for the marketplace in terms of use cases that will deliver early value. Set out agreed key performance indicators (KPIs) around these goals and priorities, so that marketplace success can be tracked. The KPIs should cover many dimensions, such as the number, quality, and take-up of data products available through the data marketplace; the number of AI innovation pilot projects launched using the data marketplace; and marketplace user adoption rates.

2. User needs must drive the design. To ensure successful adoption, data marketplaces must address the pain points of data users and publishers and deliver a seamless user experience. Getting this right means working in partnership with functional and business teams from the start to map out their key objectives and the factors that presently prevent them from achieving them. The marketplace must solve these challenges while also offering an intuitive, automated self-service user experience at the point of data discovery and access. This might include features such as Gen-AI-enabled smart searches, persona-based user recommendations, and guided user journeys.

3. Modular approach. Organizations have many different starting points determined by the extent to which they have already modernized their data environment, e.g., by developing new data platforms, data catalogs and reusable data products. The aim should be to adopt a modular, API-driven architecture that leverages the firm’s existing capabilities, fills any gaps, and can be extended and scaled over time. The modular approach to building data marketplaces can accommodate a range of ambitions — from augmenting an existing data catalog with data access workflows, to building a full-blown AI-enabled marketplace that completely transforms the data discovery and access environment.

4. Trust by design. Financial institutions need to ensure compliance with evolving data rules, including privacy, consent and localization rules across the APAC region, as well as new AI guidelines. They must also build trust among users of the data marketplace, e.g., by ensuring that the data they access is fit for purpose. Meanwhile, data owners need to know that access permissions are controlled and that data cannot be misused. Data marketplaces must therefore embed clear standards by design in areas such as regulatory compliance, governance, baseline data quality and associated checks, policy-based access controls, data masking, and well-controlled underlying workflows. There should be clarity over where data comes from, who owns it and how it should be used. User confidence can be increased through mechanisms such as trust scores, lineage tracking, end-to-end traceability, audit logs, SLAs and automated procedures — as well as through the clear delineation of roles and accountability.

5. Foster data sharing. Data marketplaces are about changing how people work and innovate. The effort begins with designing the new data marketplace around user needs. However, it should continue by training staff to use the marketplace, encouraging them to overcome entrenched data practices, and tracking marketplace adoption. One of the great advantages of a well-designed data marketplace is that it brings data users and owners together. This gives them a degree of social visibility with opportunities to find out what works best — and easy-to-use channels through which business domains, data scientists and engineers can collaborate and share ideas. 

Successful CDOs have moved beyond simply mitigating data risks and are now shaping corporate direction by ensuring that businesses and functions can access actionable intelligence. Modern data marketplaces offer a key capability in this regard, helping unite the CDO’s data risk management and value-creation functions.

Data marketplaces do not have to be Big Bang investments, nor do they depend on institutions having already developed a large set of data products. Instead, data marketplaces designed according to these five principles can adopt a phased approach to expanding data access, integrating data use into business workflows, and fostering a data-sharing culture across the enterprise.

The views and opinions expressed in this article are those of the author and do not necessarily reflect those of CDOTrends. Image credit: iStockphoto/Ahmad Bilal