We send transaction data late, and sometimes the report table is already refreshed. How do you handle late‑arriving data in the reporting pipeline?

To handle late‑arriving data, I would implement a two‑stage approach. First, ingest the raw data into a staging area with a timestamp or watermark column that records the actual transaction time. Second, schedule the report refresh to run after a buffer period (e.g., 30 minutes) to allow late data to arrive. If data arrives after the buffer, trigger a re‑run of the affected report or use incremental refresh logic that updates only the impacted rows. In a streaming context, use a time‑window aggregation that can be recomputed when late data is detected. Additionally, maintain a log of late arrivals so that stakeholders are aware of potential data gaps. This strategy balances timely reporting with data completeness.