Who Dares Define Metadata?
After purchasing a shirt at a store, you’re given a receipt that tells you what you bought, when you bought it, the cashier that served you and how you paid. Moreover, the shirt’s tag will tell you where it was made, the material it is made of, how to care for it and so on. Both the receipt and tag are sources of key information for the customer: one to confirm the transaction and the other to provide the details of the shirt. We can understand metadata in the same way.
Metadata has often been described as “data about data”, but that only tells half the story. Metadata is much more than listing a bunch of features; it revolves around the context of data – the “what, where, who, why, when and how”. It’s the data detective that acts as a silent witness to all the details that matter.
Metadata management, therefore, is the organisation of the context of data.
What to Manage?
There is growing consensus around three types of metadata. And depending on who you ask, different types are more or less important.
Descriptive metadata allows you to discover what the data is about and if it’s relevant to you. It includes information such as title, abstract and creator or even media such as youtube videos where data scientists explain the data. We often encourage new metadata management projects to start with descriptive metadata, as it provides a high-level map to understand the data landscape. Managed descriptive metadata can then be used to ‘zoom in’ on the datasets to find out more.
Administrative metadata gives you the context of the data. So once you’ve discovered it you can decide if you’ll actually be able to use it. It includes information about who is responsible for the data, access rights, provenance and the format. For those managing the data, it gives them the information they need to administer it effectively.
Structural metadata allows you to understand how the data is organised. If you’ve found a dataset you think you can access, the structural metadata allows you to dig into the data and understand how to use it. If you are looking at a database, the structural metadata would include the tables, the columns and data types in each column. Managing structural data is invaluable as many datasets will be unintelligible without this information.
Why Does It Matter?
Trust. Trusted data can move mountains for organisations. Without transparency, key stakeholders cannot make decisions that allow them to meet their objectives and move the needle forward. Untrustworthy data curbs an organisation’s ability to make critical decisions because it cannot trust the insights that inform those decisions.
If we imagine that a manager is given a list of all patients in their region, metadata asks: “Does the list include current patients or patients that visited a clinic 15 years ago? Can the patient have multiple addresses?” and so on. If you’re trying to contact a patient with sensitive and critical information, this information can be crucial. Metadata helps leaders avoid mistakes made on the basis of general assumptions.
Efficiency. Locating and using data assets easily is one of the main purposes of metadata management. If limited metadata exists, projects incur additional overheads. Where data already exists, it takes longer to find, understand and access data. Where new data collections are needed, teams often spend a lot of additional time ‘reinventing the wheel’ rather than using common models or interrogating existing datasets. Being able to locate your data and understand the practices applied leads to faster product and project delivery.
Innovation. Often large complex organisations have a huge amount of under-utilised data. Metadata improves transparency, allowing innovators to spend less time searching for data and more time converting data opportunities.
Find out how effective metadata management moved mountains for mammoth organisations here.