System of Record, System of Reference, Golden Records, and other confusing terms in Master Data Management.
It can get frustrating when trying to understand the subtle differences between various IT concepts. It gets worse when words are not used intuitively in naming these concepts! Well, if this dilemma did not exist, there would not be any need for the many articles that define and differentiate concepts. But it gets worst when the articles are too high-level or high-sounding and you need to have five tabs open in order to understand one term. So, let’s hope that this article is your article of reference for the terms explained in it!
**********************
If you are reading this, you probably know what Master Data is. Just in case otherwise,
Master Data is a category of data that provides context for transactions and events.
It is data about the nouns of an organisation. E.g: If product B was sold to customer C on Feb 2, 2019, via the e-commerce channel, the ‘things’ participating in that transaction are the customer and the product. Hence, data about the customer and product — which will most likely be reused in other transactions are part of the company Master Data. For more on Data Taxonomy, you can check Malcolm Chisholm’s article.
In Master Data Management, the aim is to control Master Data identifiers and values and ensure availability and access to accurate and current values across various systems or domains. This is what ‘we’ mean when it is said that the organisation is developing a single 360° view of an entity ee.g: the customer. To do this, Master Data values from various ‘Systems of Record’ are resolved into a ‘System of Reference’ which is supposed to be a ‘Trusted Source’ that provides users with ‘Golden Records’.
If you’re asking what all that means, let’s get to the details.
In an organisation, various information systems exist that generate, process and/or store data about various entities (referring to Master Data), operations and transactions. We classify them into two: System(s) of Record, and System(s) of Reference.
A system of record for an entity is the authoritative system where data about the entity is created/captured and maintained.
E.g: The sales team might have a system that facilitates its operations and this system will most likely process and store customer and product data. The finance department may have its own system for billing customers which is not only based on sales but other offerings & subscriptions. However, the sales-team system may be the originating source of customer data from which customer data is passed to other systems where the data may be replicated, further processed/enriched, etc. Hence, the sales-team system is the system of record.
A system of reference for an entity is the authoritative system that is expected to contain correct, complete and current data about an entity which can be used for transactional, operational or analytical purposes.
A system can be designated as the system of reference based on certain considerations or efforts that confirm the reliability of the data. For our customer data example, we may define the finance-team system as the system of reference because it contains the most accurate and current data about customers; or we may create a new system which contains the best version of the customer data from the finance and sales teams systems.
This best version is another name for Golden Record. According to whatis(tech target), a golden record is (at least supposed to be) the “single, well-defined version of all the data entities in an organizational ecosystem…..The golden record encompasses all the data in every system of record within a particular organization”. With respect to ‘single well-defined version’, it is clear and I agree as well. But this sentence makes me a bit confused about what a golden record is: is it a collection of records or a single record? From further explanation in the article, a golden record is portrayed as a system, and if that is correct, then golden record = system of reference. But this is not consistent with other articles and the Data Management Body of Knowledge which is my system of reference for Master Data (if Medium had emojis I sure would have inserted one here).
A golden record is a single record that provides correct and complete data about an instance of the entity. Hence, a system of reference for an entity should be a collection of golden records.
Ideally, we should end this article here. But the dilemma is that you may also hear/read terms like source system of record, source of record, source of truth, single source of truth, system of origin, and others I am yet to hear/read!
System of Record = Source System of Record = System of Origin = System of Entry.
System of Reference is supposed to be = Source of Truth and is supposed to be = the Single Source of Truth.
Single Source of Truth can also be a Data Architecture and Modelling constraint in which every data element is stored exactly once across the organisation.
The article has officially ended here. The next three paragraphs on Source of Record and Source of Truth are for people like me who often like to hear the other gist.
********************************
Sometimes, Source of Record and Source of Truth may be used in other contexts beyond Master Data Management.
A source of record is a system or simple documentation like a list that specifies ‘the expected state’ for an entity or subject. It is often the approved and formal system that contains data about a subject. E.g: The list of products (Master Data context) and the list of enterprise software in the organisation are sources of record. A common problem with the source of record is a lack of ownership and maintenance. This means that when no individual or department is given the right of ownership and responsibility of maintenance over the data, the system becomes out-of-date and users stop consulting it.
A source of truth is the system or documentation that gives ‘the real state’ about a subject or entity. Very often, the data is an incomplete version. Continuing with the products example from the previous paragraph, the various systems in the sales and manufacturing department may be sources of truth that each provide correct but perhaps incomplete data about the products currently being offered by the company. In the source of record, a product that has been discontinued by the company may still exist as an active product. Between various sources of truth, there may be differences in the number of products and the specified attributes for the same products.
Beyond the context of systems, source of record and source of truth may apply to organisation behaviour. For every question or decision, there are often various colleagues or managers that can provide the needed information, and there is often one colleague or manager that is most appropriate to consult. Given the same situation but different staff, the persons consulted may vary but the most appropriate person to be consulted may be same across the board. This is a classic situation of the source of record vs the source of truth.
