Sometimes it can be helpful to randomise your tasks in an unmoderated study to ensure the order of your tasks doesn't influence your results. However, there may still be instances where you want to keep some of the tasks grouped in a specific order, e.g. the two tasks need to be completed in sequence or you want to ask some follow up questions from the previous tasks. The "Keep with previous" function allows you to do that and you can read more about it here.