The invention relates to a method for timely discovery of telephone traffic system problems through telephone traffic platform data. The method comprises the following steps: (1) constructing a database, re-classifying calling platform data, and determining four basic datasheets and four auxiliary datasheets for a calling platform; (2) refining datasheet nodes, that is, refining data nodes in the four basic sheets into 63 nodes; (3) determining key research nodes, that is, finally determining 23 nodes as key nodes by researching on relationships among the 63 nodes; and (4) carrying out formula calculation through data obtained through association among the 23 key nodes to respectively obtain a loss rate, a manual answering rate and an abandonment rate, and associating formula calculation results with abnormal phenomena, platform services, equipment and data to actively find out the telephone traffic system problems. The method provided by the invention has the advantages that the concept of big data is systematically used, abnormality warning information closely associated with onsite business is captured in the first place through real-time data analysis, and warning information release achieves active operation and maintenance.