For the last two years I have been working on sound practical processes to govern data. That means creating an inventory of data by harvesting available metadata, creating metadata, and running unattended rules to ensure the metadata and data are high quality. Our team has been taking the work done in the data lake ecosystem and applying it to master data management (MDM) processes. This work has joined corporate thinking to technology, creating an MDM ecosystem to support our requirements.
The MDM technology industry is a mix of successful mature solutions and rising young solutions
Executives want to grow the organization at a rapid pace and out shine our competitors. Management thinks we need a fancy new MDM technology. Folks on the delivery teams talk about the design of the technology, and state that we need a person to model data, do deep data quality analysis, map data, ingest data, transform data, test data, manage the effort, and that we need many more people to do all the required coding to customize the solution. But what everyone agrees we need are: good technology; a small, formidable team; and re-usable processes to create and maintain high quality data.
The MDM technology industry is a mix of successful mature solutions and rising young solutions. Many of the mature solutions have an approach that a single proprietary data model with an embedded data lifecycle management algorithm will meet all organizational needs. The rising young solutions have a fresh perspective that advocates not using your internal infrastructure to manage your master data but to host your data to a cloud environment. Then rather than have a traditionally “proper” data model– which takes too much time to maintain– you can create data relationships on the fly and graph your organization’s way to growth. The one thing all these solutions agree on is that master data needs a ‘special’ environment that provides data lifecycle management processes to meet organizational requirements.
There is no silver bullet that can solve for master data management. The rub is that there is never enough money, enough time or enough people to do everything. Large organizations often have complex systems built over many years, and they cannot simply jump to a non-relational graph database. To that end, growth strategies will not be successful with only an on-premise, “single version of truth” data model either. The technologies and corporate teams need to meet in the middle with an ecosystem of solutions to support both the rapid growth mindset, as well as the need to keep the lights on.
In 2018, the path to follow is a practical approach. It seems simplistic as I describe it to company executives, peers, and industry partners. First, work to understand where all MDM data comes from and where it goes. Second, meet with stakeholders to understand critical data that drives engagement across channels. Third, understand organizational metrics and notify data producers where low data quality impacts our metrics. Fourth, for the critical master data, be relentless in driving sound, practical processes with a small, formidable team that drives the data to the highest quality to support growth and drive efficiency gain.
Finally, truly understand your organization’s requirements and be open to new ideas. Maybe a mix of on-premise and cloud solutions will meet your requirements, or maybe every master data model change does not have to go through a full data modelling review cycle. Can your teams test and learn through graphing or artificial intelligence? One thing I know for sure is that simply adding technology or people will not solve all MDM problems. New business strategies demand an ecosystem of solutions that meet specific needs.