How to choose the optimal ARIMA model

[Niema]

hi
I am working on seasonal data , first i used seasonal differencing, the correlogram shows 2 spikes for both ACFs and P ACFs , does this means i should use ARIMA(2,0,2)(0,1,0)
I did and all the coefficients are not significant.
plus there is a spike in the ACF in the 12 lag does this mean i have to use SMA?
Thanks alot :)