Signed in as:
filler@godaddy.com
Signed in as:
filler@godaddy.com
Integrating Power Apps into Power BI can be... fiddly. You can't create an app in tablet dimensions. Sometimes changes to fields don't propagate through. Sometimes the browser window doesn't open from Power BI Desktop.
When coaching my colleagues on this topic, my first tip is normally - patience. But once we're past that, I always stress that the interface they build in Power Apps should be clean, simple and neutral. Users will be seeing the app surrounded by the UI that they have added in Power BI, as well as the taskbars, panes and other clutter that exists in the Power BI Service.
So the name of the game is to design something that doesn't add to that clutter, and instead smooths the user journey as much as possible.
This app sits within a report containing profit and loss data by business category. Users select their area of accountability on one page and then drillthrough, via buttons, to the app screen. With their filters set, the app is populated with high-level metrics and KPIs and is ready for use!
So what is the app for? With the 'facts' that our metrics illustrate, our users need to create binding commitments that will form their business plan that will be reviewed regularly (with the Power BI report bringing through fresh data so we can track the impact of those commitments).
To do this, users can add up to 5 focus areas and unlimited actions within those areas. Actions can be RAG rated and viewed specifically or in totality. When complete, the business plan is saved and store in a database.
The app leverages the Power Apps PDF experimental functionality to generate a printable PDF which flexes to the amount of focus areas and actions created. This is a huge time saving for business users who were having to maintain manual PowerPoint documents, pulling data from various reports, and also for executive users to see what commitments have been made by their reports.
One usage for Power BI-embedded apps that I increasingly see is as a middle man between datasets and data science models. The model needs input parameters, that can either be completely user generated or derived from some DAX calculations or aggregates from report data.
In the above app, I created a custom connector to trigger jobs in Databricks (essentially acting as a cleaner wrapper for the API) that passed the parameters across. The user is none the wiser, but they do receive a nice email notifying them that the model has completed and the dataset refreshed and there are results to view.
Other apps allow users to review outputs from the model and take action, such as accepting or rejecting, and having those decisions propagate into other systems via APIs or data extracts.