Information Mastering Models

Extracted from the public Notion original. Notion record created 2019-12-11. Last edited 2020-01-23.

"We have an MDM solution but it's not really true MDM"

Too often "data" professionals take it for granted that they know the target solution. More often still, anything that they didn't themselves implement is somehow not even a true solution and is never considered best practice.

Unfortauntely, they are probably right. But also, we get to design our organisation however we want, but we need to do this in a rational way that doesn't just ocntinuously throw away the old only to replace with something new but basically the same.

Data management professions tend to reserve the term "mastering" for a view specific sub-set of the types of ways that data might be brough together and made available to others. It's prefered that we have a broad definition of "mastering" and recognise how many types we have, the diffrent types we would like, and where we need to do some work.

You can imagine a range of "mastering" options:

1. Just bringing data together from different sources on to the one phyiscal storage with no effort to ensure the data is integrated, or verified, or even defined

1. Bringing data relating to the same data domain or subject area into the same phyical location, but also taking the effort to bring this data into a unified data model - in this instance a data model that is the sum of the data sources

1. Bringing data relating to the same data domain or subject area into a data model designed to be not just the sum of the data sources but a model that has been designed with at least some level of top-down, strategic perspective that incorporates not only the data we have, but the data we would like to have, and the types of data structures that would truly support our target operating model, customer experiences, and risk management

There are also other dimentions of mastering that you might incorporate, such as one or more of the following:

1. Matching real-world entities (such as people) if they appear in different data sets without confirmed common identifiers

1. Applying agreed data quality rules that attempt to understand the quality of the data

1. Accessabilty improvements that ensure definitions, structure, and publication approaches are incorporate to ensure that the data can be discovered and assessed for suitabilty for new uses