The current configuration is global per entity and applies the same task count to every execution above a single threshold, regardless of the actual volume or legal entity.For example, with a threshold of 200 records and a task count of 8:An import of 20,000 products creates 8 tasks of roughly 2,500 records each, where parallel processing is likely beneficial.An import of 2,000 products also creates 8 tasks, now with only around 250 records each. This consumes the same batch/AOS capacity while potentially adding more scheduling and partitioning overhead than performance benefit.A more adaptive design could support either:Multiple volume bands per entity, such as 2 tasks above 400 records, 4 tasks above 1,500, and 8 tasks above 10,000; orA target records-per-task value combined with a maximum task count.This would allow DMF to calculate an appropriate degree of parallelism for each execution rather than treating every import above the threshold identically