Large-scale multi-objective optimization problems (LSMOPs) can be difficult to solve due to the complexity of the search space and the limited computer resources available. The decision variable grouping technique is a promising method for addressing this challenge, but existing techniques often fail to guarantee solution quality and speed. In this paper, we propose a conserved sequence (CS) grouping strategy for LSMOPs that overcomes these limitations. Our approach begins by dividing decision variables into conserved and non-conserved sequences using a decision variable analysis method based on the conserved sequence. We then apply specific optimization strategies to each sequence: the conserved sequence is updated directly along the directions of its parental evolution process, while the non- conserved sequence is optimized using an embedded multi- objective evolutionary algorithm (MOEA). Experimental results on a range of benchmark test problems demonstrate that the pro- posed CS performs competitively compared to eight state-of-the- art algorithms, with a negligible computational cost compared to two representative grouping methods. The effectiveness of our approach is also demonstrated on a real-world problem.