In backtracking search, exactly one successor is generated at a time during the search process. This means that the algorithm generates successors incrementally, exploring one successor node before backtracking and trying another if needed. This approach helps to use less memory since only one successor is expanded at each step rather than generating all successors simultaneously

. This behavior contrasts with other search algorithms that might generate all successors of a node at once. Backtracking's strategy of generating one successor at a time aligns with its depth-first nature and the need to backtrack upon hitting dead-ends or constraints. Thus, the number of successors generated in backtracking search at each step is:

• One (1) successor per node expansion