Data Governance: The Art and Science of Turning Data into an Asset
In a decade, data has become a strategic asset valued in the same way as a customer base, patents, brands, and other intangible assets of any company, whether born before or after the era of digital, data, and AI. To achieve this, the data itself has undergone two major transformations. Initially descriptive thanks to data visualization, then predictive thanks to machine learning. Now, it is becoming prescriptive through decision automation, realizing the virtuous cycle long promised by digitalization: data, predictions, decisions, new data and so on. Companies that have adopted both transformations have become more agile, adapt and sometimes anticipate changes in their customers and their environment. But transforming data into predictions and then into decisions is not automatic. They all had to implement data governance — the art and science of organizing, unifying, and enriching data so that it becomes actionable by data scientists and business analysts.
Big bang of data
Since McKinsey’s report on big data in May 2011, we have entered an era where almost everything we produce is designed to be digitized in order to be able to generate data. It has become a strategic asset in any company — not just Google, Amazon, Meta, Apple, and Microsoft, which have paved the way. Recent transformational projects, including the Metaverse, want to transform all human activity into a digital economy with data as the central element of this new economy. But before getting there, the data itself has gone from the status of a purely technical resource, stored in databases and data warehouses, to the stage of an asset valued among the other immaterial assets of any company, whether it was born before or after the era of digital, data, and AI.
From data to prediction
Barely ten years ago, data was an IT resource supporting businesses and functions and as such, it was generally managed by the IT department, whose mission was to build the architecture data, choose a database supplier and design the applications linking the databases to the needs of internal users, the majority of which are business analysts and financial controllers. These applications were primarily dashboards, allowing the business to get an idea of how it was doing against its goals.
Then came the first revolutionary transformation in the sense that it allows companies to be able to project themselves into the future. Dashboards have been enhanced by predictive analytics whose scope of analysis is no longer just about what happened in the past, but also what might happen in the future if the company doesn’t make significant changes in its corporate strategy. In general, predictive analytics focusing on the future covers most of what data science is today, with use cases in advertising, marketing, sales, and customer relationship management, all sectors combined ranging from retail to luxury and beauty to finance and insurance.
From prediction to prescription
Since then, a second transformation has been underway, first in the banking, insurance, and health sectors in the USA and China, but it will certainly cross the oceans to land in Europe, with its promises, its realities, and its new regulations. It consists of transforming predictive analytics into operational decisions.
The promise of this second transformation is to create a virtuous circle where not only is data analyzed, but that analysis is transformed into decisions and actions that generate new data, and so on. Reporting and predictive analytics are now complemented by prescriptive analytics. For example, the technology developed by Sparkling Logic, a Californian company, is used by major American financial institutions to automate the credit process for individuals, based on data provided by the individual and risk scores calculated by predictive models.
But, contrary to what we might think, transforming data from a technical resource to a strategic asset is not easy. Indeed, all the companies dream of being like Google, Amazon, Facebook, Uber, and Airbnb but they were not all born with data and machine learning in their DNA. As a result, their data is not directly actionable to derive a strategic competitive advantage as is the case for data-native companies. This situation gave rise to data governance, a new discipline combining business know-how and technical know-how to bridge the gap between companies born before and those born after the big bang of the data era.
Data governance
Aiming to use all the data available in a company to extract predictions and then decisions requires a new order of work. It is not enough to bring together all the company’s data in a data platform or data marketplace for the data to be transformed into knowledge, forecasts, and decisions. Indeed, all the data do not have the same age, the same structure, the same format, the same quantity, the same quality, and especially the same usefulness. If an attribute of a product or a customer is important for a division, it is not automatically important for another division, even if they are both parts of the same organization. Each division has its own vision of the product and of the customer. In the luxury sector for example, a dress, a bag, or a jewel, although it is a unique object, is seen through different attributes depending on the databases where this same object is stored.
Like any acronym, there is no consensus on the definition of data governance and it should not be confused with data management or data quality management. For my part, I define data governance as the organization, processes, and tools to be put in place so that the data is ready to be activated by the models and the algorithms data scientists would have built for the business. Otherwise, data science would not deliver its promises — transform data into value.
Successful data governance initiatives always had business owners in addition to data architects, data modelers, data engineers, and data scientists. In an architecture model represented in the form of levels, where the highest level represents the needs of the business and the lowest level that of the technical resources, data governance level is located below data science level and above data management level.
From a practical point of view, data governance integrates, unifies, and harmonizes data originally stored in various sources, ranging from data warehouses and databases to shared and local files. It is also at the level of data governance that industry-specific and regional regulations are implemented. The multitude of regulations that are flourishing all over the world, sometimes with regulations by state in the same country, makes data governance even more complex and expertise even more sought after. Businesses now have to integrate restrictions such as the impossibility of using data outside the territory where they were collected, or even of using recommendation algorithms without the ability to explain their outcomes. Dreaming of a “one size fits all” global data platform is no longer relevant. From now on, pragmatism prevails — one platform per continent, country or product line.
Conclusion
In less than a decade, data has gone from being a resource to evaluate the performance of a company to an asset used to predict the future of this one. Recently, data has become an asset used to automate and improve decisions. These two rapid transformations were made possible thanks to an awareness of the strategic side of data governance, without which there is no data intelligence. This transforms companies into continuous learning organizations where data helps identify opportunities, machine learning transforms this data into knowledge, and decision automation transforms this knowledge into action, closing the virtuous circle that data promises.
As for its future, no expert is able to predict it exactly, but one thing is certain is that data is interfering in all economic activities to the point that it has become ubiquitous as an asset that investment and venture capital firms value in the same way as the customer base, patents, trademarks, and other intangible assets. Even anonymized as increasingly required by international and sectoral regulations, data will retain its now strategic value. This explains why new entrants prefer to capture data first, even if it means losing money, but for a much greater return on investment.
Wrap-up
- For a decade, data is used to not only evaluate the performance of a business but to predict its future. Now, it is used to transform these predictions into decisions and actions.
- But transforming data into predictions and then into decisions is not automatic. Consolidating all data into one data platform helps, but is not enough on its own.
- A new order of work called data governance is required without which data science will not deliver on its promises — turn data into value.
- Data governance is what transforms data from a technical resource managed by IT to a strategic asset managed by the business.
About the author
Consultant and interim executive, specialized in data strategy, data governance, and data platforms for the retail, luxury, finance, and insurance sectors.
An earlier version of this article appeared in French in INfluencia and in English on VentureBeat site.
