What might be evolving around Master Data Management?
There are many ways that master data and its management is evolving. Many external factors are influencing MDM, some of the main ones we’re looking to explore within this article are AI, how technology is advancing and new capabilities around data mesh and fabric.
In the past, MDM was often seen as a centralised activity, the rise of federated governance models (such as in data mesh) calls for a more nuanced approach. Each business domain may find value in shaping its own definition of master data, tailored to its specific challenges and requirements. For example, a customer from a marketing point of view may be different to a customer from supply chain or finance point of view, while still sharing some characteristics. With the right technology, governance and business process, this is a perfectly achievable scenario. This shift from a one-size-fits-all approach to a more customised understanding underlines a key principle in modern data management: adaptability is crucial.
The sources and methods used to define master data are evolving, with an increasing focus on integrating external sources to create a more efficient, richer, and more comprehensive dataset. This shift recognises the growing importance of external insights in strengthening the internal data ecosystem. By incorporating third party sources such as Dun & Bradstreet, GS1, external data pools, or leveraging data syndication technology organisations can enhance data accuracy and efficiency. Additionally, the accelerating use of AI for content generation reduces the time needed to produce high-quality, reliable datasets while boosting confidence in the integrity of the data.
Artificial Intelligence (AI) and Machine Learning (ML) are transforming numerous fields, often in ways that go unnoticed. Their influence extends across industries, revolutionising processes and opening doors to efficiencies that were previously unimaginable. At the same time, the methodology behind defining master data and associated data models is shifting towards AI and machine learning-driven processes. These technologies offer a more dynamic, value-driven approach to data analysis, moving away from the traditional, often time-consuming, attribute-by-attribute analysis. AI and GenAI are now accelerating business-as-usual (BAU) data processes, not only in data analysis but also in validation use cases. For example, in Customer MDM, identifying duplicate customers - a time-consuming manual task - can now be significantly sped up with AI driven recommendations or application of matching logic. Additionally, AI can help us to validate large volumes of data rapidly, bringing a new ability to check data accuracy with minimal human intervention. For example, flagging unrealistic values, or identifying anomalies, to improve accuracy of these processes and reduce the risk of errors.
Lastly, the role of metadata – essentially, master data about your master data – is coming to the fore. Its management is becoming increasingly integrated with the master data entities themselves within MDM solutions, underscoring metadata’s critical role in understanding and leveraging data effectively. When metadata is well-managed and accurately defined, it significantly enhances the effectiveness of AI applications and improves the quality of the insights generated. These transformations in the definition, sourcing, and management of master data are not just trends but necessities, ensuring that businesses remain agile and effective in a data-centric world.
All of this has made clear that MDM is being impacted by a variety of disciplines across data management. We’re going to take a look at some practical advice for approaching some of the main influencing factors below.
What are your next best steps for a modern MDM approach?
The Dos and Don’ts for AI & Generative AI
- Do: Use technology (including AI) in order to support your team/improve productivity
- Don't: Fall into a trap of thinking that technology alone- for example active metadata or generative AI - can automatically generate, manage, or provide the full depth needed for your Master Data and its model.
The use of AI and active metadata present huge opportunities within the MDM world to decrease time to value enabling organisations to make use of actual data to build semantic models at a pace and scale not possible through human analysis alone. This can also bring businesses confidence that design decisions made are built on real data and assumptions will hold up even across multiple data sets.
Technology however (at least for now) simply can't automate human decision making, policy definitions, and real-world applications and the context of data. The reality is also that these processes are still limited by the quality - or lack of - of the underlying source master and transactional data being analysed. They require clerical review and augmenting with human knowledge and data specialisation.
Only when ML / AI is used in conjunction with business leader expertise, and users understand how to interpret its findings can real opportunities for acceleration be achieved. Often, the greatest value lies in areas where labour-intensive processes can be simplified or automated. Many organisations are already implementing straightforward use cases, such as generating product description copy for business review, defining master data models, or utilising AI for data translation.
It’s essential to select use cases that align with your organisation’s goals, while also being mindful of associated risks and any relevant regulatory requirements.
The Dos and Don’ts for Fabric / AI
- Do: Start utilising MDM models and updating them as business needs and data changes.
- Don’t: Assume that your MDM model can be designed and built once and stay relevant.
We're seeing data and models evolving at a faster pace than ever, driven by the rapid growth in data volumes and the increasing demand for data across the business. These developments demand a proactive and adaptable approach to data governance, enabling the effective monitoring and support of changes to master data models.
The key to keeping up is through integration and automation. Approaches like data fabric and metadata model integration across systems, including MDM, are great at boosting efficiency. They move us away from a reliance on manual model governance, towards a more efficient approach. This can look like technology not just supporting, but actively suggesting updates to the MDM data model. For instance, it might propose adding new fields or crafting new validation rules based on emerging data trends.
The future might even hold the possibility of automating some of these adjustments, further streamlining the evolution of the models themselves. This approach aligns perfectly with the needs of rapidly evolving data landscapes, ensuring our data management strategies are as dynamic and forward-looking as the data they govern.
The Dos and Don’ts for Technology Features
- Do: Consider what capabilities may be needed on top of your MDM/PIM solution in a modern data ecosystem.
- Don't: Automatically think you need to scrap your existing MDM or PIM solution in favour of a vendor claiming to be aligned to modern data concepts or capabilities. Effective MDM cannot be achieved through technology alone and this fact is as true as ever, and regardless of your architecture of choice.
Careful consideration is needed about the functionality you need your MDM or PIM solution to bring (but this shouldn’t be anything new). While it’s true that technology alone will not solve the challenges of a modern data ecosystem, it remains an essential enabler. Not all of these technologies will be critical – but some may become more so, depending on your business data strategy and architecture of choice. If you have existing MDM or PIM technology, consider its current capabilities and pay attention to the vendor’s roadmap in some of these emerging areas.
Things to consider for PIM / MDM tech capabilities that might be needed in a modern data ecosystem.
- Does it have the capability to make metadata readily available, exportable, and integrable?
- Does it provide capabilities for integration with third party data sources, and existing generative AI platforms?
- Can it manage visibility privileges in order to provide different views for different roles, and even entirely independent data models to facilitate business domain-specific use.
- Will it handle automation of data model or configuration changes used with caution.
- Observe technology marketplace in general, consider whether key use cases or functionality can be achieved through lighter applications rather than a single platform.
The Dos and Don’ts of Data Mesh
- Do: Encourage data silos to be unified so master data can be centrally accessed.
- Don't: Assume that Master Data Management is still synonymous with providing a single version of the truth, nor does this have to contradict data mesh architecture principles.
In certain contexts and data domains, the notion of having a single representation of an entity, such as a customer, is a genuine need – a customer cannot logically exist more than once in reality. This understanding shapes our approach to Master Data Management (MDM).
The essence here is ensuring data connectivity across different business areas, thereby removing the necessity for each domain to have its distinct view of a customer with separate data models, attributes, and values. This strategy significantly lessens the need to force every business sector into a uniform model.
Instead, a more targeted approach might be more suitable, focusing on mastering data for specific business domains. This involves considering the capabilities of your MDM platform, including aspects like permission management, data model maintenance processes, and the adaptability of the underlying data model.
A singular MDM platform can be provisioned and maintained by IT whilst permitting federated data ownership within particular business domains. This arrangement creates a balance between comprehensive oversight and domain-specific data management, ensuring that the representation of crucial entities like customers remains consistent and accurate throughout the organisation.
Final thoughts
Modern MDM isn't about scrapping everything and starting fresh, it's about evolving what you already have to meet today's challenges. As businesses grow and technology advances, MDM needs to keep pace, ensuring that your data is not just managed but leveraged to its full potential.
So, what does this mean for you? It means taking a hard look at your current MDM landscape and asking the tough questions: Is it adaptable? Can it integrate with the latest technologies? Is it helping your business stay agile and responsive? Is it providing the data you need, and in a reliable manner? The goal isn’t to throw out what you have but to make it better and evolve with the times.
If you’re ready to get started, our Modern MDM guide is here to help. It offers practical steps to modernise your MDM while building on your existing strengths.