Following up on a recent post about data silos I have been thinking (and remembering) a bit about the idea that one company can have all master data stored in a single master data hub.
Supply Chain Musings
If you for example look at a manufacturer the procurement of raw materials is of course an important business process.
Besides purchasing raw materials the manufacturer also buys machinery, spare parts for the machinery and maintenance services for the machinery.
Like everyone else the manufacturer also buys office supplies – including rare stuff as data quality tools and master data management consultancy.
If you look at the vendor table in such a company the number of “supporting suppliers” are much higher than the number of the essential suppliers of raw materials. The business processes, data structures and data quality metrics for on-boarding and maintaining supplier data and product data are “same same but very different” for these groups of suppliers and the product data involved.
Supply Chain Centric Selling
I remember at one client in manufacturing a bi-function in procurement was selling bi-products from the production to a completely different audience than the customers for the finished products. They had a wonderful multi-domain data silo for that.
Hierarchical Customer Relations
A manufacturer may have a golden business rule saying that all sales of finished products go through channel partners. That will typically mean a modest number of customers in the basic definition being someone who pays you. Here you typically need a complex data structure and advanced workflows for business-to-business (B2B) customer relationship management.
Your channel partners will then have customers being either consumers (B2B2C) or business users within a wider range of companies. I have noticed an increasing interest in keeping some kind of track of the interaction with end users of your products, and I guess embracing social media will only add to that trend. The business processes, data structures and data quality metrics for doing that are “same same but very different” from your basic customer relationship management.
Conclusion
The above musings are revolved around manufacturing companies, but I have met similar ranges of primary and secondary constructs related to master data management in all other industry verticals.
So, can all master data in a given company be handled in a single master data hub?
I think it’s possible, but it has to be an extremely flexible hub either having a lot of different built-in functionality or being open for integration with external services.
Liliendahl on Data Quality 
I agree with you that it is possible. Infact I have seen it in action. The task to solve is to agree where to cut.
To me, all things in the real world have names. This is at it’s most abstract level of course and does not relly help much – but it is a fact that humans have become accustomed to give everything names.
Before we assign a name we have to decide what type of thing it is that we are naming. The type from there on can govern what data we expect to gather for the thing in question.
The database I referred to earlier could actually store persons and producst in the same structure – and it works wonders to the flexibility. Not only to the database structure but to the processes using the data and to the integration of many disparate structures.
Of course before the data was stored it was standardized to fit our MDM structure. And that was by far the most difficult part of the work.
But, yes, it is possible.
Arnt-Erik, thanks for the comment. Abstraction can get us a long way for sure.