TV Ads are still considered as one of the most expensive cost in a particular promotional activity, hence it needs to be effective in accordance to its viewer neuro-psychological behavior. Electroencephalogram (EEG) can capture brain activity and inform a person’s brain behavior while watching a particular video ad. This research proposed neuropsychological identification in real time every three second captured by an EEG wireless and then processed using Fast Fourier Transform (FFT) and Support Vector Machine (SVM) with three classes of: 1) interested, 2) less interested, and 3) not interested. The subjects were asked to fill out the interest questionnaire after recording to determine the class from the training data. Extraction using the FFT was performed by changing the frequency in 4-40 Hz, which contain alpha, beta, gamma and theta waves. FFT of all frames, was used as an input identification system using non-linear SVM by finding the best hyperplane that distinguished each class. The identification of three classes was done in storied. First, the researchers separated the interested and other classes. Then the second SVM was separated from the other classes into two: the less interested and the not interested. The research was conducted by identifying 9000 training data using 30 subjects for each 3 trials, 10 segments, watching 10 ads, video. The results showed 91.5% accuracy of training data, while testing against 4500 new data from 15 subjects received 71% accuracy with SVM gamma of 100. With 0.3 compute time, it was not significantly toward 10 seconds identification. Therefore, the system can be used in real time.