How to apply a rolling Periodic Date Filter to PA Indicators (bypassing the Bucket Group limitation)
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
2 hours ago
Hello Community,
I am running into a roadblock trying to apply a periodic date filter to Platform Analytics (PA) report on a dashboard, and I am hoping someone has a cleaner implementation approach.
My Use Case / Setup
We are tracking custom AI tool usage on the Scrum Task table. We have an "AI Use" choice field (ChatGPT, Gemini, NowAssist, Other, Not Used, etc.), and if "Other" is selected, a single-line text field captures the specific tool.
We built a dashboard where we need to filter the data by a periodic selection (e.g., Today, Yesterday, Last Week, Last Month).
The Good: This works perfectly for our standard Data Visualizations using an out-of-the-box Date Interactive Filter.
The Bad: We have two specific widgets (Total AI Tools Used and Total AI Tools Not Used) that must be built as PA Automated Indicators to accurately calculate the count distinct over time. As expected, the standard Date Interactive Filter does not natively filter these indicator-based reports.
The Attempted Workaround (and the Pitfall)
To solve this, I researched using a PA Script to calculate the record's age and mapping it to a Bucket Group breakdown (e.g., Today: 0-1 days, Last Week: 2-8 days, Last Month: 8-31 days).
However, because PA buckets are mutually exclusive, they create distinct, non-overlapping slices of time. This completely breaks the user experience for rolling date periods.
Example of the impact:
Task A (Created Today) -> Drops into the Today bucket (0-1 days).
Task B (Created 4 days ago) -> Drops into the Last Week bucket (2-8 days).
Task C (Created 15 days ago) -> Drops into the Last Month bucket (8-31 days).
If a leadership user selects "Last Month" on the dashboard filter, they logically expect to see a rolling 30-day view containing Tasks A, B, and C. Instead, the report strictly shows Task C (excluding the more recent tasks because they sit in different buckets).
My Question
Can anyone suggest a better implementation approach for handling rolling periodic interactive filters on PA Indicator-based reports?
Thanks!
