In this paper, we propose a low-complexity hybrid analog-digital signal detector for uplink multiuser massive multiple-input multiple-output (MIMO) systems. Specially, both the hardware cost and computation load can be reduced. The analog part consists of phase shifters only which are tuned by the idea of equal-gain combining (EGC) for each data stream. Thus, no additional optimization is required, and the number of radio-frequency (RF) chains decreases at the same time. In the digital part, the Gauss-Seidel (GS) iteration with decision is then used to extract the transmitted symbols from all uplink users. Simulation results show that our proposed hybrid processing method provides similar bit-error rate (BER) performance as compared with the fully-digital minimum mean-squared error (MMSE)/zero-forcing (ZF) detection, and is even better for some cases. When considering the complexity reduction, it is an attractive way for giving a good compromise between the error rate and complexity.