Master data: Why best practice is far away and what we can do about it

The gap in terms of data quality between the global elite companies and the bunch is widening. During the evolving AI era, a state-of-the-art master data approach is even more crucial. It is not too late, but you should start now.

Companies around the world look at global hyperscalers and digital-first companies to get inspiration about AI. Strangely enough, they tend to simultaneously ignore the data discipline and foundational master data culture enabling the exponential scalability among these high-tech companies. Thousands of companies are still far away from best practices regarding Master Data Management (MDM). The digital transformation of traditional companies offers huge potential to solve this problem at the roots, but it is an issue for executive management and not (only) for the CIO office.

Huge and costly problem

Everyone digging just slightly into the problems with master data agrees about the implications. Lost revenue, increased operational costs, compliance violations, and reputation damage are among others. Gartner estimates annual costs of $12,9 M on average among enterprises because of poor master data.

This should be serious enough for every company to take a firm position and invest accordingly to solve this problem. However, a cascade of primarily organizational issues creates obstacles to the right approach, and in my view, technical issues are secondary to management barriers on the way to robust master data.

Root cause with many faces

If an address of a customer is incorrect, the delivery will fail, and the problem must be fixed with a manual effort in customer service – and hopefully with no additional consequences. At the same time a flawed product description leads to inventory problems in several locations, and a management decision is being postponed because of a lack of consistent data. Each of these problems are probably solved by an extraordinary effort, where they occur, but they all derive from the same root cause: Poor master data quality.

We have a discrepancy between the ownership of the master data issue and the economy of master data problems. You would probably look at the CIO or the CFO office to find responsibility for the framework, culture, and processes of MDM. If this issue is not solved accordingly, the costs are spread out in various functions and approached as smaller (or bigger) here-and-now exceptions.

Foundational investments

In 2008 chief economist at Google, Hal Varian, said that the sexiest job in the next ten years will be statistician. In light of the current boom in the AI area, the quote had a strong visionary quality. I would suggest that the master data architect should be awarded the same title for the next ten years.

The value of successful MDM – deeply rooted in decision-making, technical approach, and company culture – can not be underestimated. But it requires C-level attention. The benefits gained from a foundational approach in the CIO office will be big but hard to measure, as it will translate to better operations with fewer mistakes everywhere.

AI will probably be integrated more and more into operations and decision-making in most industries and will fuel the issues with master data. What we do not want is to produce, expand, and spread errors. But at the same time, AI will be integrated into frameworks and tools designed to strengthen master data quality, and the intelligent software has the potential to address the master data challenges more effectively.

Hundreds of business rules

Issues with master data have a long track record because of its position between the tech domain and the business functions. It is complex in many organizations that are not born digitally and therefore carry a history of multiple systems, several generations, and therefore diverse formats, platforms, and methodologies.

Many IT professionals have historically seemed frustrated about even establishing an overview of maybe hundreds of business rules regulating enterprise processes. Now intelligent software and specialized consultants can “sniff” it up, categorize, and describe the business process rules for you.

What can we do?

The way forward is cross-functional and it requires a sign-off from C-level. I would recommend a value-centric approach to MDM as opposed to a focus on cost reduction. Best-in-class MDM free organizational resources and make room for growth and innovation.

A set of principles should guide an all-encompassing MDM effort. Simplification is one of them, as the reduction of complexity leads to focusing on the essential. Another key principle is automation. It is truly logical to relieve humans and instead have software taking care of hundreds of business rules.

Companies need to accept the fact that master data plays a role in end-to-end business processes. This leads to the necessity of establishing a governance framework. Who does what? We need to support employees at an operational level, and we need to protect master data from local and siloed changes.

Another important principle is pro-activity. Do not build your MDM effort on a philosophy of reactive response and regard MDM as a motor for correction. Instead have an approach focused on the elimination of errors in the first place, which can be done by robust control and guidance in master data maintenance processes .

And when?

These years AI is fueling an already ongoing digital transformation in most companies. Leaving older on-premise ERP systems behind and embracing standardized cloud-based ERP platforms is part of the journey. Many companies need to have a serious look at data and data quality as part of the migration process.

Having taken this big step of cleaning up and arriving at a clean sheet, a stringent data foundation is a perfect starting point for a proactive MDM culture. Now is the time to put an end to pollution and discrepancies in data by establishing a governance and technical framework across the enterprise. Fostering a culture of data quality requires ongoing training and awareness for employees to support and understand the automated functionality.

You may want to use several software solutions in different areas of your system landscape in your company-wide MDM approach but the principles should reach from end-to-end, and the goal should be consistent master data of high overall data quality as key enabling components to AI-driven innovation.