Introduction
Before starting a new buzzword, we need to revisit an old one: MDM. Master data management (MDM) is a term used for the last few decades to help with the growing challenge of having a "single view" of entities across organisations, such as customers, businesses, and products. It became clear as organisations started to use multiple systems, maintain data in multiple places (aka silos), merge with new businesses and getting data from external sources. The evolution of big data, IoT and AI have only made that challenge of single view even bigger. Despite the continuous proliferation of data and data management tools in the past decade, and the challenges accompanied with that, most MDM solutions and initiatives failed, and failed miserably.
In a few recent engagements, customers showed how they try to avoid talking about MDM as a solution, or somehow manage to give (or ask) for different names for the solution, and that's due to its history in under-delivering on its promises and goals. I can't blame them. The past 20+ years have proved that MDM solutions from "leading" global vendors were unreliable, not scalable, and simply don't do a good job- they under deliver. And surprisingly, with all sorts of advancements we've seen in many areas in data management, these solutions managed to stay unchanged, using the same methods and techniques, year after year, to tackle an ever-existing challenge.
MDM has fundamentally failed (and continues to) because it:
- Promotes the idea of "single view"
- Deals with data and data issues as a first-class citizen
- Continues to impose old techniques and rigid models to deal with data sourcing, processing, and serving
In this 3-part blog series, I'll be covering each of these points. Below, I'll dive into the first reason on that list, and that's the tendency to promote the idea of a "single view" MDM solution.
There is no such thing as a "Single View"
Or at least, it shouldn't be thought of that way. For organisations to be data driven and provide a true culture of "data democratisation", they need to provide their users with the right data whenever they need it. Each user's/team's requirements are different, and the way they define or see an entity is different from one another.
Let's consider an organisation that has multiple systems for storing customer data for different purposes. This organisation has three departments that need access to customer information: fraud, billing, and marketing.
For obvious reasons, the fraud department would want to have a flexible view of customer data in order to investigate potential fraudulent activity. They would be interested in seeing all relevant information about a customer, such as emails, phone numbers, driving license number and addresses.
On the other hand, the marketing department would only be interested in the current details of a customer in order to send them, for instance, a gift. They are likely to be restricted from seeing certain information about the customer, such as their driving license number.
Finally, the billing department would be interested in having a explicit view of all accounts that a customer may have in order to accurately bill them for the services they are using or subscribed to. For example, the same customer could have a business account and personal account, and each should be billed separately.
The Importance of Having a Contextual View
Therefore, a "single" view isn't viable anymore (if ever) and could potentially be wrong or even impose governance risk. Organisations grow, and so do their business units, vision, goals and data assets. That means a flexible view is becoming more vital than ever, and a traditional static "single view", or "golden record" doesn't fit for most purposes, with the exception of a few use-cases. To deal with multiple use-cases using this traditional approach means generating multiple copies, where each copy serves a group, user, or use-case. This doesn’t only imply multiple physical copies, but also multiple efforts, costs associated with that (i.e. of infrastructure, of projects, of resources, of security overhead of storage), time, maintenance and so on.
In short, a “single view” doesn’t fit for purpose anymore, and our perspective should be changed in understanding the difference, and the importance of having a dynamic Contextual View that's readily available when needed, on the spot.
Next: Rethinking MDM and Our Approach to It | Part 2/3: Data is not a first-class citizen
