The churn pattern is more relevant than the churn rate since agentless replication takes in data. When a file is written multiple times, the rate has little effect. In the next cycle, however, a pattern in which any other sector is written produces a lot of churns. You can let the data fold as much as necessary before scheduling the next cycle if you want to transfer the least volume of data possible.