Forum Discussion
Creating a *useful* Power BI Data Dictionary for Viewers?
My perception is Data Dictionaries usually fail (not just for Power BI, in general with tech). Maybe because
- Too complex and/or out of date for end users
- Too much hassle for developers to update (plus end users aren't using it anyway...viscious cycle)
I see different ways of auto-ish creating a data dictionary in Power BI, but while they could dazzle, they don't seem useful.
- In model view, populating the Descriptions - nice, clear spot, lives direclty in the report.
- Descriptions are only visible to builders though, not end-viewers. So to expose them:
- XMLA Endpoints (if you have Fabric).
- Using the new-ish built in views. These two stand out:
- info.view.measures() for measure definitions
- info.view.columns() for column information
But even those descriptions don't seem that useful to an end user.
- Column descriptions could be useful: If the visual the viewer sees, uses the exact column name
- Measures descriptions and definitions coudl be useful: If the visual the viewer sees, uses the exact same measure name (which usually they don't - the model measure likely has a verbose name, while the visual employes a shorter, possibly context-specific name)
So maybe, one really needs a data dictionary that references the actual visuals the viewer sees. So how to do that:
- A text box within the report is simplest: Easy to enter, doesn't make a report on Live Connection switch to a composite direct query. Can't have slicers though.
- Home > Enter Data: That creates a proper table one could use slicers on. But more of a pain to update than a text box, and does mean a live connection report has to swtich to a composite direct query model
- A linked Excel file: Great in most ways, but a bummer that it lives outside the report
- Images pasted in from an Excel file that lives somwhere
I have no good conclusion, thus asking for input.
Anyone had luck, with a sustained, useful data dictionary for Power BI report viewers?
5 Replies
- lbendlinSuper User
Create a MVP (minimum viable product)
Introduce it to your audience.
Observe them interacting with the report
Note down all the terms they are struggling with
Create a data dictionary for these terms
Rinse and repeat - adjust the list according to the changes in your audience.