The Data Quality Conundrum

Today’s challenging regulatory environment and the growing cost and complexity of information acquisition and management are compelling asset managers to evaluate their data management strategies. Many firms focus on resolving isolated issues and fail to consider other associated malignancies or possibilities of downstream corruption or inconsistencies. Unrecognized consequences due to patching isolated problems often result in a lack of confidence in data by business users who increasingly recognize data as a key corporate asset that drives competitive advantage. Lacking a macro perspective or defining a more holistic strategy that transcends departmental lines and silo systems can result in a firm's consuming valuable resources and squandering money from typically stretched IT budgets. But, it isn’t easy.

In the past IT was responsible for data management and quality initiatives. This is shifting, and the business side is becoming increasingly the steward for data quality. Though this migration from technology to the business accepting responsibility is philosophically well underway, the implementation is not yet complete. In the meantime IT continues the struggle in understanding nuances in the data and to referee competing organizational demands concerning that data. At the same time business users are growing increasingly frustrated with inaccurate results, processing delays and recurring restatements. This cross-organizational collective disillusionment has instilled renewed vigor in asset management firms to actively seek solutions to resolve information and data management issues.

Quality is never an accident; it is always the result of intelligent effort.
      - John Ruskin
        (19th century writer, art and social critic)


Rules of Engagement
Data management encompasses processes, technology, and people. Adopting strong data governance and assigning stewardship frequently obliges firms to adopt new rules and approaches while fostering a partnership across the business units and IT. Ongoing support and commitment by senior management is crucial to success and will ensure that resources are allocated, budgets are refined and cross-departmental cooperation is attained. Traditional roles and responsibilities are metamorphosing to create a new paradigm for aligning and harmonizing data across the enterprise and systems. Opinions can vary on the best practices, and firms are typically fine-tuning prospective models to most effectively support their businesses.

Two examples of approaches for data quality management include implementing either an upstream or a downstream data quality strategy. Upstream strategies focus on the data source and benefit every application that may draw from the source. These systems are the parents for all data flowing through the organizations. For upstream strategies to succeed, organizations need to ensure clear ownership, consensus and the support of the staff across all applications. This can become challenging in large organizations in which, historically, data has been under the ownership of lines of business or due to mergers and acquisitions, multiple data sources, disparate systems, cultures, and processes are being integrated. Less than half of the respondents to a recent Cutter survey have built formal logical and physical data models or implemented firm-wide data dictionaries that are maintained regularly for all their data. In fact Cutter noticed that there are not many formal procedures or tools in use at the enterprise level. Most firms still have a downstream strategy that involves applying data quality at the downstream applications or in divisional data warehouses. Benefits with the downstream model are realized by users of the applications or data residing in the warehouse. Generally downstream strategies are easier and less expensive to implement.

Whether the firm chooses the upstream or the downstream strategy, three elements are critical to the success of the program: 1) senior management buy-in and on-going support, 2) well publicized and controlled initiative scope and 3) defined and tracked metrics to evaluate ongoing success or craft adjustments over time.


Anchors Aweigh?
Not only is ownership of data quality up for debate, so are the acceptable levels of error or even the data definitions. Is a Beta or PE that is outside of x standard deviations of an index considered “garbage” that requires review? And what index should be used in the comparison? Is a supranational security with no country code incorrect? How many decimals should be calculated?

Tolerances and definitions associated with the quality of data vary by department as well as across different firms. Defining what is acceptable and gaining consensus is critical to a successful data management strategy. Generally accepted key metrics for tracking quality include accuracy/precision, completeness, reliability, and availability. Though commonly regarded as the primary measures, organizations vary in opinion on what are appropriate levels and weights concerning the various measures in the process. Some firms believe it is more relevant to be consistent, others seek absolutely correct data and yet others are satisfied with the best available with recognition that data is never perfect and may need further adjustment as time passes. Whatever your firms primary objective and level of satisfaction, the focus is having consensus on the target and implementing the technology, process, and people skills to support that target.


Full Speed Ahead
Given the complexity and nuances of asset management data and the impact incorrect data can have on the overall success of an organization, asset managers are more actively pursuing and engaging external vendor solutions. The primary drivers are having access to newer technologies and providing the ability to enhance speed, streamline processing, proactively uncover problems, and actively engage the business units with user-friendly and intuitive tools.

Shifts in thinking within asset manager organizations have coincided with significant changes in the data quality vendor landscape. ETL, Database and BI vendors are acquiring niche data quality vendors as part of their overall strategy to support the data capture and dissemination process. Reference systems, too, are building out their data quality capabilities. But many niche players still exist and offer exceptional capabilities, which are system and architecture agnostic. The vendors of all of the above solutions are not just increasingly interested in providing data cleansing functions, but are also becoming increasingly focused on providing robust functionality to support process management. Many solutions offer the convenience of accessing any data source whether relational, dimensional, flat file or XML; some even support unstructured data and in real-time; some offer only profiling (uncovering gaps or issues in the current data); while others offer full matching, linking and enrichment; some offer in-memory cleansing to alleviate the writing of results to a warehouse or other structure to perform the cleansing. Such features exemplify the significant progress being made in this discipline and should cause organizations to pause before they undertake yet another effort to plug a hole in their data process or fix another data problem with an isolated patch. Maybe it’s time to look at the bigger picture.

Watch
Recognizing the significant issues and complexity in resolving asset management data issues has a growing number of vendors, both those historically focused on the asset management industry and those that are not, actively involved in helping solve the conundrum. Differences in these systems not only include depth and breadth of functionality, but also differences on the core models for evaluating and cleaning data, and positioning where the solution fits in the overall logical architecture and workflow a firm adopts.

On May 16th in New York and May 22nd in London, CutterResearch™ will be presenting a detailed review of data quality systems and processes at The Technology Council™ meetings. The presentation will also be an opportunity to hear how large global investment firms are tackling these challenges.