Vendor Accent

Data Quality: the first metric that matters for PM success

Ram Guduru talks about the benefits of Performance Management and points out the importance of data quality

Performance management (PM) solutions, combined with associated best-practices and business processes, are rapidly becoming recognized as the best vehicles for the most strategic use of an organization’s information. In fact, Hackett Group research illustrates that companies with world-class enterprise performance management generate 2.4 times three-year equity market returns, including stock price increase and dividends, of typical companies in their industry.

Through disciplined PM practices, companies can align their operational and financial information, internal processes, and strategic goals to build competitive advantage, increase return on investment, and drive superior business results. PM technology combines data from a multitude of sources and transforms it into actionable performance information that provides a single, consistent, accurate view of corporate processes and performance. This process ensures that reliable performance information reaches the right people in the right way at the right time. With PM, business users throughout the enterprise can consistently make better decisions around three key performance-related questions:

• How are we doing?– Organizations can measure and monitor performance with scorecards and dashboards that track key metrics.
• Why?–Reporting and analysis capabilities let organizations uncover the reasons behind good and bad performance by exploring the data, gaining context, understanding trends and spotting anomalies.
• What should we be doing?–Through plans, budgets, and forecasts, a business can set and share realistic and reliable views of the future.

As the appetite for consistent, reliable performance information continues to grow, accompanied by the demands of increasingly stringent compliance legislation, many organizations are recognizing the indisputable importance of data quality. If performance management solutions are to be effective, it is imperative they be built on a foundation of high-quality data that delivers a single version of operational and financial performance. This is a non-negotiable requirement and it is the first metric that is fundamental to successful performance management deployments.

Armed with the knowledge that PM based on quality data will result in greater commercial opportunities, forward-looking organizations are addressing the issue of data quality as an essential component of their performance management implementations. They are seeking to ensure that they have the best possible quality data on which to base critical business decisions.

The key dimensions of data quality

Given its critical importance, the logical question that arises is, “What exactly is data quality and how does one measure it?” Good data quality is defined in a number of different ways, but ultimately it is about the data meeting the needs of the information consumer. To measure data quality we need a well defined set of metrics that we can use to set targets for data quality and measure conformance to those targets. Organizations quite often begin their data quality process with three or four of the metrics or dimensions below and then add dimensions from this list or define new ones, such as timeliness, to track as their processes mature. Developed as a guide, the following six dimensions can help business personnel achieve a fuller picture of data quality and better understand how to optimize it within their enterprise. They also provide a common language that enables business and IT professionals to work together to deliver the highest levels of data quality.

• Completeness: Does the organization have all of the relevant data? Are there empty or default values in fields? What elements are missing or unusable and how will their absence affect the organization’s PM initiative.
• Conformity: This deals with the format of raw data and content related issues such as incorrect format within the field. For example a name prefix in the customer name field or noise around telephone numbers. In addition, businesses must assess what data values are stored in non-standard formats. For example, does a part number (which should contain only digits) include spurious alphabetical characters?
• Consistency: What data values return conflicting information? For instance, in a record, you might have a currency field for the United States with the currency represented in Euros. Or a US address might have a Canadian alpha-numeric postal code.
• Duplication: Are there records that are repeated and skewing the data? There should always be a unique ID associated with each record, but often when organizations look more closely at the raw content of other fields, there is a high degree of probability of duplicates. For example, they need to check if there are aliases that should be aggregated (e.g. “International Business Machines” and “IBM”)?
• Integrity: What data is missing important relationship linkages? Fuzzy matching can be used to identify records which should be linked to each other.
• Accuracy: This concerns comparing data with a reference source. For e.g. using a lookup table for an exact match to see what data is incorrect or out of date. In this scenario, a purchase date cannot come before a birth date.

A business-focused approach

To support these six dimensions and address the critical aspects of an organization’s data quality, companies should look for an open, platform neutral architecture that addresses their need for solution standardization. Data quality deployments should focus on the business user’s role in ensuring data quality. Data across the full breadth of the enterprise must also be addressed to achieve success.

Furthermore, data quality and performance management implementations should operate in a virtuous cycle that combines people, processes and technology in an iterative process of continuous quality improvement. This linkage ensures that the appropriate members of the organization’s business team take ownership of their data and collaborate with IT to provide effective business rules for transforming this data into consistent, actionable information across the organization, ultimately enabling higher performance.

This type of an approach to data quality stretches beyond point solutions that call for an exclusively IT approach to address data quality. These types of approaches result in an incomplete resolution of data quality issues and do not effectively address the dimensions of data quality discussed previously. While IT is responsible for getting the raw data, they do not “own” the data contents or are necessarily aware of its context. This is the role of the business.

Business or IT: who owns data quality?

Traditionally, businesses seeking a consistent, complete view of information focused only on the information management layer of the extract, transform, and load (ETL) processes involved in converting data to actionable information. Many still do.

As organizations’ PM data increasingly comes not just from a single data warehouse but from a wide range of disparate sources, and as more organizations combine transactional, financial, and operational data to drive business decisions, the historical, IT-centric approach to data quality is no longer adequate.

Today, data quality excellence to support performance management deployments mandates an approach that combines both business and IT. While IT focuses on the data infrastructure needed to support performance management, business users must focus on the business rules that determine what information is provided, and that requires knowledge of business needs and an understanding of the language and nuances that must be brought to the table when collaborating with IT. In short, data quality in support of BI and performance management demands an integrated effort by IT and the business.

Bringing PM capabilities to bear on the issue of data quality means that ongoing data quality can be managed using data quality dashboards, scorecards and alerts to continuously improve and monitor the quality of the data that underpins business performance.

This approach lets business users monitor key data quality metrics (data quality dimensions), just as they do other performance-related metrics. Consequently, business users are engaged in the process of improving and owning the quality of their information. This “buy-in” helps facilitate the on-going process of data quality and enables the more rapid adoption of data quality and performance management initiatives.

Better quality data for better performance

Because high quality data is central to the strategic business needs of today’s organizations, data quality has become a priority on the corporate agenda.

By combining data quality technology with their performance management technology and initiatives, organizations can ensure the quality of the data on which their performance depends is monitored and continually improved.

The author is Regional Director, South Asia, Cognos ram.guduru@cognos.com