Whether an alignment rate can be considered "good" depends on the source and library prep kit used for the sample, so it can be difficult to assign a hard cutoff. That said, you might want to look more closely at samples with less than 70% alignment. 

Keep in mind that the number of reads that end up aligned, rather than the percentage of aligned reads, is often more important. A low read count aligned to the genome can call into question the validity of any downstream results.

A Sankey diagram in the Expression Count (STAR) pipeline provides sample alignment metrics.