This article is the final installment in a multi-part series on data governance by Arun Krishnamurthy, Managing Director at Grandview Analytics, about how asset and investment managers can establish a robust data governance practice and a culture of data. Explore earlier articles: Introduction to data governance, Assessing and advancing maturity, and Choosing your data governance operating model for more.
There are six distinct components to a robust data governance practice. Depending on your organization and the data governance operating model you use, one or many different teams could be involved. To ensure consistency, repeatability, and scalability, each team executing and implementing data governance components will adhere to the guidelines and guard rails of the data governance policy and framework. In this final installment, we discuss how these essential components come together in practice by sharing a common use case and describing the considerations and interdependencies.
In context: a common use case
Let us consider the use case of a buy-side asset manager in an early stage of data governance maturity that desires higher confidence in their security master data set. For this use case, let us assume the asset manager’s data platform consumes security data from market data providers like Bloomberg and IDC, while also consuming security data from the front-office order management and back-office accounting engine. Each of these data sets are consumed, matched using matching rules, and then a master internal security record is created using hierarchy rules for the data from each source. The result is a data platform that has an internal security record as the “single version of the truth” comprised of data from each source and associated with the security record from each processing platform and market data source.
Next, we will revisit the six components of the data governance framework in more detail, with the above use case as context, to appreciate how data stewardship assignments align to support a strong framework and business objectives.
1. Data stewardship:
Data stewardship involves managing your organization’s data to ensure its quality, reliability, timeliness, access, and usage. Data stewards “own” the data operations for a domain, focusing on data content, lineage, access, data catalog and usage (both current and future), while data custodians focus on data management tools and technology.
Let us consider the security (issue) data use case described above. The data steward team is responsible for:
- Ensuring the appropriate data quality rules to validate security data content are defined, implemented, and executed, and security data from each source platform is correct and consistent.
- Maintaining the data catalog that describes the source of the data and critical data elements (those security attributes required for business, operational, and data processing) across the processing platforms and market data providers.
- Creating and maintaining the lineage of the security data from creation through the investment lifecycle, including whether the security data attributes are from a market data provider or an internal processing platform and how the security data was processed (matched, mastered and created/updated).
- Documenting the current usage of security data and having a proactive view of future use cases. For example, is your organization planning to source additional security data for a different/new asset class because a new product is being introduced?
- Defining the classification of data and determining who should have access to the data. Some derived attributes may be classified as “confidential” and thus restricted from the broader audience. The data steward team owns this decision.
The data custodian team, in partnership with the data steward team, implements the appropriate technology according to the capabilities defined and required by the data steward team. This could be a data quality tool, a data catalog tool, or a workflow tool to define, request, and grant access to data.
2. Metadata management:
Metadata management creates a shared understanding of your organization’s data, making it easier to use data consistently across departments. It provides answers to critical business questions such as:
- What does this data represent for our business?
- Where did it originate and how has it been used?
- How does this data support our business processes?
By maintaining clear definitions, lineage, and usage guidelines, metadata management ensures that all stakeholders can confidently use data to make informed decisions. Data stewards maintain this information in a data catalog, which serves as a reference point for technology, operations, and business teams.
Continuing with the security data set example, the data steward would utilize the data catalog tool to document and store all the metadata. This includes business meanings for attributes, matching rules for security data from each source, hierarchy mastering rules to create the internal security master record or your organization’s common definition of security type, and the mapping of security type data from source systems to the standard/common/internal values of your organization. In addition, lineage information is captured in the data catalog tool.
In mature data catalogs, the business processes may also be defined in the data catalog along with the data attribute and how that attribute is used. Based on your organization’s data culture maturity and use cases, the catalog could be maintained in an Excel spreadsheet or, more likely, using modern data catalog tools such as Alation, Colllibra, Ataccama, or data.world. Regardless of choice, the data steward team must own the maintenance and change control processes for the data catalog.
3. Data quality management:
Data quality management guarantees that your data is accurate, consistent, and available when needed. This is critical not just from a technical perspective but for business decision-making as well. For instance, high-quality data reduces the risk of errors in reporting, leading to more reliable insights for portfolio management and client reporting. Business-specific data quality rules ensure that data aligns with operational requirements and business goals. Data stewards, in collaboration with business and technology teams, work together to monitor and maintain data quality, ensuring issues are addressed swiftly and without business disruption.
The data steward team defines rules to validate the timely delivery and content of the data. For example, there could be a data quality rule that ensures that the security type provided in the data set is valid given the expected set of data values for that security type. Data quality rules can be executed before the data set is processed, during the processing of the data set, or after the data set is processed. The optimal location for data quality rule execution depends on the use case. There are several feature-rich data quality platforms such as DQLabs, Talend, or Ataccama, and your firm’s data custodian team should implement a fit-for-purpose solution.
4. Data issue management:
Data issue management tracks and resolves data exceptions. This workflow provides transparency about the state of issues, their categorization, impact, resolution, root cause, and preventive measures. This process enhances trust in the data.
In our use case, let us say someone from a portfolio management team disagrees with the security type value of a mutual fund. While the mutual fund is categorized as equity, they believe that the mutual fund should be categorized as fixed income, as the fund primarily holds debt instruments. The portfolio management team could utilize the data issue workflow tool to challenge the assigned value. As the data issue is being investigated, researched, and resolved, the status can be reported back via the data issue management workflow tool. In this specific case, the security data operations team may report back that the mutual fund is categorized as equity since the processing performed on the fund and the characteristics of the fund are like an equity.
This transparency into the reasoning, reporting, and tracking of root cause analysis and resolution provides users with a higher level of understanding and trust in the data. The original submitter of the challenge may accept the reasoning, or they can raise the issue with the data steward, but the fact that it is simple to submit a challenge, track its progress, get the reasoning behind the original decision, and then accept or follow up with a change request helps build trust and credibility. The data steward team “owns” the process and content of the data issue management workflow tool while the data custodian team “owns” the platform from a technology perspective.
5. Data access, privacy, and compliance:
Managing data access involves ensuring data is accessible but controlled. For example, compliance with regulations like GDPR (General Data Protection Regulation) is achieved by managing what personal data is held, its purpose, processing, and access.
For example, there may be a Quant team in your organization that requires access to security data from one of your data providers to perform back testing. The Quant team would submit a request for access to all the required security data for back testing to the data steward team, who would review and fulfill the request according to its documented data access policies. It’s possible that the security data from the provider may not be available to the entire organization due to licensing, so access may have to be restricted, or require the license to be modified so that access to the data can be provided. Having a consistent and controlled mechanism by which access is requested, reviewed, and granted is especially helpful with restricted data sets like personally identifiable information (PII data) or position and transaction data.
While the data steward team “owns” the process, decision and final resolution, the data custodian and information security teams provide the appropriate technology tools and platform to make this possible.
6. Data architecture, model, schema, and integration:
A well-designed data architecture ensures that your data flows efficiently across the organization, reducing errors and improving the quality of analytics and decision-making. By aligning the business and conceptual data models with your organization’s needs, data architecture enables seamless integration between systems, improves operational efficiency, reduces complexity for technology teams, and accelerates the delivery of insights for the business. A robust architecture means fewer errors, faster access to data, and more accurate analysis—allowing the business to make data-driven decisions with confidence.
Continuing with our example around security data, your organization would define the architecture of how security data is created, maintained, used, and flows in the various technical, operational, and business processes. While the ultimate responsibility for creating the data architecture may be with the data custodian and data engineering teams, business processing and architecture decisions will require input from data stewards, operations teams, and front office teams; data modeling decisions will require input from the data architecture team. In this collaborative process, some questions the teams may consider could be:
- Do securities get created in trading systems and then flow to the accounting engines?
- Or: Are the same securities created separately at both platforms with a well-defined method to correctly link the trading system security to the accounting platform security?
A pragmatic approach to data governance
Defining, aligning, and executing data governance framework components will drive consistency, reliability, effectiveness, and scalability for your organization and build trust in the quality of your data among the data consumers. As you continue to move along the data maturity curve, your firm can fully leverage data as an asset to increase positive business outcomes and create a robust data culture.
Establishing a data governance framework and program from scratch is a meaningful and significant organizational change effort and can be overwhelming. We recommend a pragmatic approach: start with a small, well-defined set of data, establish the data governance framework and policies, implement the processes, measure outcomes and value, and then continue to iterate and improve.
Are you ready to strengthen your data governance practice and unlock the full potential of your data? Whether your firm is in the preliminary stages or looking to optimize an existing framework, Grandview Analytics is here to support your data governance journey. If you have any comments or questions, feel free to reach out to Arun Krishnamurthy.
