It's the nature of evolutionary algorithms. Normally, we run multiple trials.

GA is a stochastic and population-based algorithm. The variations in results is due to the use of random numbers at the mutation and the cross-over stages. However, if you are coding the algorithm yourself, you can control the behaviour of the random variables in these stages.

due to the mutation process

Any EA starts with random generation of initial population. Initial Population is made up of many random solution of the problem at hand. Further as mentioned above random search such as crossover and mutation in case of GA leads to random answers everytime. EAs are mainly used for solving problems where even a suboptimal real time/polynomial time solution is acceptable.