The Anchors of Trust in Data Analytics
An exploration of some of the critical questions and challenges emerging around trust in data and analytics. The four anchors of trust that will shape public confidence in D&A in the age of the analytical enterprise are highlighted.
By Brad Fisher, Partner, U.S. Leader for Data & Analytics, KPMG.
Complex analytics form the foundation of many important decisions that affect us as individuals, businesses and societies. With so much riding on the output of data and analytics, questions remain about the trust we place in data, the analytics and the controls that influence decision making. How do we know that a result is ’right’ or that automated decisions are doing ‘the right thing’? What does ‘right’ mean? Few organizations think about ’trust’ beyond data accuracy and security. We see a heightened need to focus on trust and believe that it will become an increasingly important factor in the value proposition of data and analytics. Those who can manage trust will move forward with confidence, both in their decision making and their customer relationships.
How can the analytical enterprise assure trust throughout the analytics lifecycle amid an increasingly complex ecosystem? While there is not yet a standardized framework for assuring trusted analytics, it’s important for business leaders to consider what we believe are the four critical dimensions that can help assess the degree of the trust gap.
The first dimension relates to quality. Are the management processes in place around analytics and data good enough? This includes practices related to the accuracy, provenance and ‘freshness’ of data. In many organizations, leaders raise questions around data ‘lineage’ (i.e. where the data originated and what process it took to arrive at the analytics machine). Data consistency and completeness are central components to data quality.
Following quality is the concept of accepted use. Organizations and analytics experts need to clearly understand if the intended analytical approach is appropriate to the context in which it is being used. For example, crime statistics could be used as a proxy for economic vibrancy in a specific geographic area, but only if the right statistics are being leveraged and in the right context. Organizations will also need to ensure the way they manipulate the data is appropriate and defensible. Knowing when and how to appropriately apply data and analytics to various scenarios is key.
Once accepted use is determined, accuracy and utility of prediction should be examined. Executives and analytics experts will need to ensure that the analytics ‘work’ to achieve their intended purpose, and that the predictions and insights are accurate and reflect reality.
Finally, there is a growing ethical dimension to analytics. Organizations need to consider whether the way they are using data is ethical and that ensuing predictions are managed with integrity. The lack of transparency and discriminatory data use (for example, if crime data is used as a proxy for race) are already major concerns. This is an area of great uncertainty and rapid change, with enormous potential reputational risk.
Trust in analytics must be addressed immediately if society, governments and consumers are to fully reap the benefits of using data and analytics. It cannot wait for regulators to step in or for a major issue to force legislators to intervene.
To read more about ‘trust’ as it applies to data and analytics, please visit: www.kpmg.com/trust.
Bio: Brad Fisher is the D&A leader for KPMG’s Innovation and Enterprise Solutions in the U.S., and a partner with more than three decades of experience providing professional services to clients in a variety of industries. Brad serves as an internal data ‘evangelist,’ working to leverage KPMG’s advanced capabilities in Big Data, predictive analytics, optimization modeling and analytics technologies to enhance the firm’s professional services.
- The Perpetual Quest for Digital Trust
- AnalyticsStreet Panel Report: Frontiers and Dangers of Analytics and Big Data
- Big Data Analytics in Hotel Industry