This paper investigates the impact of request aggregation on delay performance in group-based machine type communications (MTC). To alleviate the network congestion caused by concurrent transmissions from massive MTC devices, group-based MTC with request aggregation are shown to be one of the effective schemes. In group-based MTC, the devices are grouped into clusters with a dedicated cluster head (CH) which aggregates the requests sent by the cluster members. Hence, the amount of concurrent transmissions for connection establishment to the base station (BS) can be reduced. However, how to determine the amount of aggregated requests in one CH-to-BS transmission, namely aggregation factor, is still an open issue. To this end, we formulate an optimal aggregation problem to minimize the request delay subject to a drop ratio constraint. Based on our analytical and simulation results, we present the performance curves to relate the aggregation factor and the request delay as well as the drop ratio. Consequently, we suggest the optimal aggregation factors for MTC applications requiring ultra-low latency and ultra-reliable delivery.