Separating high value from low value KPIs in data governance efforts

Register now

(Editor's note: This is part two of a two-part series. Part one can be viewed here).

Key performance indicators for data governance help IT decision-makers determine if their business strategies are working. However, as mentioned in part one of this two-part series, all KPIs need periodic assessment to ensure they're adequate. If they aren't, adjusting the KPIs is a crucial step to take.

This second part examines some of the specific KPIs related to data governance and management that companies might track. It features three high-value KPIs and three lower-value metrics.

1. High-Value KPI: Percentage of Duplicate Records
Duplicate records could unknowingly skew results and put data governance professionals at risk of negatively impacting an organization's reputation. Moreover, the dangers of duplicate data could cause incorrect conclusions about the number of customers, sales statistics associated with a given period and more.

So, establishing a KPI about duplicate records could make a company more confident in the quality of its data, and thereby more able to rely on it for decision-making.

2. Lower-Value KPI: Number of People Who Received Data Governance Training Across All Locations
It's understandable if companies want to track how many employees take part in data governance certification days or similar kinds of training. However, if the organization in question consists of several locations across multiple states, countries or continents, a KPI that measures a type of training and does not break down the results by location is too broad.

It's better to choose that data governance training KPI and make it more specific, such as by focusing only on one branch of a company or all the locations in one state. Otherwise, it may be impossible to see which sites are meeting expectations for data governance training and which ones require improvement.

3. High-Value KPI: Number of Resolved Data Quality Issues
Data quality issues are numerous and can result from date-related inaccuracies, numerical mistakes and inconsistencies with units of measure, among other things. That's why it's advantageous for data governance experts to consider creating a KPI related to the number of resolved quality issues.

Regardless of which direction a change occurs, the outcome could be telling for an organization and its teams.

4. Lower-Value KPI: Number of Overall Data Quality Issues
It's not necessarily a bad thing for a business to know how many overall data quality problems happen in a specified span. But, the reason this could be a lower-value KPI in many organizations is that it is likely not adequately specific. In contrast, a KPI for resolved issues indicates if a company is making gains in remedying problems or not.

Looking at the number of data quality issues also becomes more problematic if a company has numerous locations. Failing to separate data quality issues into outstanding and resolved categories could also promote inaccurate presumptions about performance.

Indeed, it's best to keep the number of data quality issues as low as possible. But, it's arguably even more critical for company representatives to accurately ensure they're promptly addressing and thoroughly handling all issues. Only viewing overall issues may not represent how those problems get treated.

5. High-Value KPI: Number of Sensitive Data Locations
All companies that could potentially handle the data from European Union (EU) residents must abide by the General Data Protection Regulation (GDPR) — even if those enterprises do not have a physical presence inside the EU. The GDPR specifies how companies must treat data related to customers. Failure to comply could result in substantial fines.

A vital component of the GDPR framework is the right to be forgotten. It's also sometimes referred to as the right to erasure. In certain circumstances, individuals can make verbal or written requests to have their details deleted from a company's database. Also, companies must respond within a month.

A 2016 survey showed only 12 percent of organizations knew where all sensitive data resided in their databases. It's not hard to imagine, then, how difficult it could theoretically be for company representatives to find data associated with individuals who submit erasure requests.

The GDPR differentiates between personal data and sensitive data. The first kind encompasses things like names and phone numbers. The sensitive data umbrella, then, relates to things like information about genetics, trade union memberships and race or ethnicity.

To be clear, the GDPR's right to be forgotten extends to personal details. But, it's still useful for a company to know where it keeps sensitive data, mainly to avoid unintentionally mishandling it.

6. Lower-Value KPI: Number of Data Governance Meetings Held
It's essential for teams involved in data governance to hold periodic meetings to recognize known obstacles, celebrate achievements and provide tailored input. However, a KPI for the number of data governance meetings held during a timespan only indicates people got together for discussions. And, as anyone who has ever sat through a meeting knows, some are valuable, but others aren't.

It is instead better to create a KPI that measures how many people participate in new data governance programs, demonstrate an understanding of key concepts, etc. Companies should consider creating a KPI about whether compliance with a new or revised data governance program goes up at a satisfactory rate after its introduction.

Such a KPI could be instrumental in determining whether an organization thoroughly communicated a new program to its team members, and it's arguably much more useful than one that merely tracks the number of meetings a company has. Knowing whether sessions translate into meaningful changes within a company assists the organization in seeing how useful those gatherings are.

As this list shows, it's crucial for organizations to put a substantial amount of thought into the KPIs they create. None of the suggestions here are useless, but some have more value than others.

For reprint and licensing requests for this article, click here.