What is Metadata?
Put simply, metadata is data about data. The more information you can create and store describing a piece of information, whether it be a book, a web page or a digital asset provides humans and computers alike to identify, understand and discover particular assets quicker and easier.
There are a number of different types of metadata, and broadly fall into 2 or 3 widely accepted categories (depending on who you talk to), mainly that of Descriptive metadata (discovery and identification), Structural metadata (how the components of objects are organised) and finally Administrative metadata (data related to the administration and management of the source).
The Enterprise BI Portal leverages all 3 types as defined here. Descriptive metadata of each asset facilitates ease of discovery, Structural metadata governs how assets are grouped and what hierarchy is applied, and finally Administrative metadata provides the necessary information of what type of asset it is, when the asset was registered and so forth.
Let's look at these in a little more detail.
How does the BI Portal use Metadata?
The nucleus of the Enterprise BI Portal is asset metadata. It dictates the structure, governance, and discoverability of each individual asset registered.
1. Descriptive Metadata
The process of registering assets in the portal is one that natively forces you to capture necessary information to ensure the new asset belongs where it should and can be found easily. Basic information such as "Name" and "Description" allow you to provide brief, straight to the point information of the asset. Let's take an example of a new Marketing report we want to add.
In this case, the;
- Name "Customer Profitability" and,
- description (a brief outline of what the report is about),
- the author (me in this case),
- The type of asset (Power BI report)
- A link to the report
- Any tags I choose to assign to the report such as "Customer Profitability" "Profitable" "Customer"
- Any SMEs I would like to assign, in this case I'll make it myself, and Adam.
All of this new data that we created, the "metadata" is now related to the new report we have registered. This means, I can look for this report at a later date, by searching on any one these aspects defined.
2. Structural Metadata
By definition, structural metadata refers to the manner in which assets are organised. In the portal, this is a direct reference to the "Business Area (or Department) and the "Subject Area" assigned and what we refer to as "guided navigation".
Navigation is provided by the way in which you would order your content. Usually by business area or department. In the example above you can see that the Customer Profitability Sample report is assigned to 3 departments "Purchasing", "Leads" and "Marketing" as the content of the report is relevant to the teams present in all of these areas. If you navigate to "Marketing" you will be presented with this report. If however, you navigate to a business area that has not been assigned to the asset, then of course, this report will not be visible.
3. Administrative Metadata
Referring to technical metadata, Administrative metadata includes information such as when the asset was published, the type of asset it has been classified as, and also the URL of the asset.
All assets registered and / or published in the Portal also have administrative metadata associated with them, and can be searched by such. In the example below, I have searched for the term "Power BI"
As you can see, a number of reports have been returned, all with term Power BI assigned to them.
Metadata Best Practices for the Enterprise BI Portal
Given the importance of metadata and the role it plays in the Portal, here are some handy hints and best practices we recommend to all of our new and existing BI Portal customers:
- Design your Enterprise taxonomy first
The portal comes with a standard taxonomy stamped in, but this may not suit every business. Get focus group together and at the very least determine the structure of your organisation, the various departments, and the relevant, common, subject areas that would fall under each department. Ultimately, the way you structure your taxonomy and translate to the Portal, will determine how your analytics assets will be governed, and discoverable.
As you can see in the example below, we have created our primary navigation as "Departments" . This is known as the "Navigation Term Set" and can accommodate multiple tiered terms underneath. Do not get carried away with the number of levels that you create. We would recommend 2 or 3 at the most.
2. Use common terms, follow logical groupings and naming conventions
You want people to use the Portal intuitively. To ensure everyone can navigate to and find the asset that they are looking for, ensure that you use common, relevant term descriptions, familiar to your business. Most business reports will be about a particular subject area, relevant to a particular department(s). If you think about how people would look for a particular report, this will be a useful guide on how you can structure the guided navigation in the portal.
3. What about tagging?
You can assign unstructured metadata to assets in the form of tags. All assets can be tagged as many times as you would like, but whichever description you deem relevant to the asset. The caveat here however, is to understand the usefulness of tags. Tags belong to individual assets, and are usually specific to the particular nuance of the asset. I would hesitate to tag a report "finance" as we already have a term set (navigation item) called "finance" that would allow me to find the report anyway. If however, the report was particular to a project called "Red Arrow" I would tag the report "Red Arrow".
This means, that if I know the name of the specific project I can search by the term "Red Arrow" and the report will be discoverable. In the situation prior, this would not be the case as there would have been no metadata related to "Red Arrow" assigned to the particular report.
4. Use Audience Targeting Sparingly!
The portal allows content to be audience targeted. A wonderful feature, and one that allows you to serve up relevant content to the right people. However, audience Targeting should be used sparingly. If you do not need to lock down assets, do not audience target terms as it will add unnecessary overhead in the management and governance of who can see what.