Anomalous or low‐quality seismic data from seismometer malfunctions or incorrect metadata can interfere with real‐time seismic processing and degrade products provided by earthquake early warning systems. Thus, it is important to monitor data quality to detect sensor failures rapidly and exclude anomalous data from processing. Here, we detect data anomalies and malfunctioning sensors by comparing instrumentally corrected waveforms between co‐located broadband and strong‐motion seismometers in Southern California Seismic Network (SCSN). We assume that a signal within the common resolution range of both sensors should have near‐identical instrumentally corrected waveforms. Specifically, two waveform consistency metrics, amplitude ratio and cross‐correlation coefficient, are evaluated. Both metrics should be for normal data; deviation from 1 in either metric indicates at least one of the sensors is producing unreliable data. We examine these two metrics for 1 yr of local earthquake records and identify 32 problematic channels out of a total of 672 in the SCSN. In addition, we show the feasibility of near‐real‐time data quality monitoring through measuring the inconsistency rate over a short period of time using anonymous large‐amplitude signals. We highlight this method because of its general detectability in a broad variety of data issues and ease of integration into real‐time monitoring systems. This method is expected to help identify malfunctioning instruments and enhance overall network data quality, which is required to lay a solid foundation for robust earthquake early warning and other real‐time seismic processing systems.