Incremental Refresh deletion

[mrbelchi]

Does the rolling window automatically remove old partitions during each refresh, or is this behavior dependent on specific conditions (Premium capacity, manual XMLA operations, etc.)? I've found conflicting information about whether old data is automatically pruned as part of the refresh process or if it requires manual partition management.

Example:

• Archive data starts 5 years before refresh

• Incrementally refresh starts 3 days before

• Refreshes daily

When data ages beyond the 5-year archive retention window, is it automatically deleted during each scheduled refresh operation, or does it require manual intervention?

• Day 1 (Sept 19, 2026): Initial refresh loads 5 years of archive data (Sept 19, 2021 onwards)

• Day 2 (Sept 20, 2026): Scheduled refresh runs

  • Does data from Sept 19, 2021 get automatically deleted since it's now older than 5 years from the current date?

  • Or does the retention window remain static from the first refresh?

Thanks

[cengizhanarslan · Accepted solution]

Old partitions are automatically removed as part of incremental refresh, no manual intervention needed.

Power BI maintains a rolling window, not a static one. On each refresh, it:

• Refreshes the incremental range (e.g., last 3 days)
• Drops partitions that fall outside the archive window

[Olufemi7]

Hello @mrbelchi,

For your scenario (archive 5 years, incremental refresh 3 days, daily refresh), Power BI automatically deletes partitions older than the retention window during each refresh. On Sept 20, 2026, data from Sept 19, 2021 is removed because it is now outside the 5-year rolling window. The retention window moves forward with each refresh; no manual intervention or XMLA operations are required

Docs: 

Configure incremental refresh and real-time data for Power BI semantic models

[mrbelchi]

Thank you for the doc link 👍