Before publishing a dataset to production, it is important to go through a checklist to ensure its quality and accuracy. Do you have your own? I do. Please, keep reading to discover more about it.
📢 To clarify, I assume that you have finished the prerequisites such as requirements gathering, modeling, and data validation. My aim is to provide recommendations for documentation that can enhance your work and improve the experience for users of your dataset. Now that everything is cleared up, let's get to it.
My checklist ✅
Hide unnecessary tables, columns, or measures to simplify data for end-users.
Create hierarchies in dimension tables and groups when necessary to further simplify the analysis experience.
When dealing with dimensions that have a large number of columns, it may be beneficial to group them into "folders" based on their functionality and with the end-user in mind. However, it is important to avoid unnecessary grouping and consider the potential inconvenience of having to navigate through multiple layers to access a specific column in a table.
Create a measure to display the date and time of the last update.
Create a measure that shows the version of the dataset.
Please write a description of up to 500 characters for your dataset, keeping in mind the limit imposed by the Power BI Service. Your description should contain a brief overview of the main objectives and functionalities of the dataset, which we refer to as BigPicture. Additionally, please provide the department and contact details of those responsible for the dataset. You may also include primary data sources and the frequency of updates if you wish, though these are optional. This description will be featured on the front page (e.g. Home or Cover page) of the .pbix file (page 1) and also in the Power BI Service's description.
Create a cover page that includes the following:
The company logo and/or department logo, and use the brand's color palette to ensure everything is in line with the brand's image.
The description you previously created.
A card displaying the measurement of the most recent update.
Another card displaying the version measurement of the dataset.
If necessary, include an additional text that grabs the reader's attention. This cover should be placed in the first position.
Delete pages that do not have a validation approach. evidence is removed.
Name the pages with a practical sense focused on validation.
Hide pages, leaving the cover visible. If necessary another, the one that common sense indicates.
To ensure a thorough data validation approach, it's best to delete pages that do not meet this goal.
When naming pages, prioritize practicality and relevance to validation: tabla name, functionality, and so on.
If needed, you can hide certain pages while keeping the cover visible. Use common sense to determine which pages to hide (or ask your colleagues).
To set up dataset approval in Power BI Service, it goes through two stages. The first stage is called "Promoted" which happens after it passes functional validation. The second stage is "Certified" which occurs after it passes the organization's quality standards review process (Governance). Once a dataset is certified, it becomes "Reliable" and can be safely distributed within the organization. This feature is specifically designed for organizational models.
Apply sensitivity labels: If your organization uses it, kindly configure the sensitivity labels.
I would like to receive your feedback, ideas, or related practices to share and work better. And if this post has helped you, it makes me happy 😁.
Cheers 🤟
Matias