An asset event can carry a small JSON object, its extra. The producer sets it while the task runs:
@task(outlets=[DAILY_SALES])
def load(rows, ds=None, outlet_events=None):
n = warehouse.insert("daily_sales", rows)
outlet_events[DAILY_SALES].extra = {"day": ds, "rows": n}
outlet_events cannot be assigned to; set .extra on its entry. yield Metadata(DAILY_SALES, {"day": ds}) (from airflow.sdk) does the same from inside the function.
The consumer asks for triggering_asset_events: a mapping from each asset to the events that started this run, oldest first. triggering_asset_events[DAILY_SALES][-1] is the latest; read .extra, .timestamp or .source_dag_run (the producing run's dag_id, run_id and logical date). This is how an asset-triggered run, which has no logical date, learns which day to process.
Keep extra small and JSON-serializable. It is stored in plain text and shown in the UI, so never put credentials in it. inlet_events[asset] reads earlier events of an asset outside an asset-triggered run.
The report reads the day from the event that started it.