True, both methods can provide statistical significance. My only point is that in general A/B testing may require less number of visitors as compared to A/B for same amount of confidence.