This article proposes a new hybrid multi-target regression and classification deep-learning-based framework for the prediction of the effectiveness of various methods with which B2B marketing can be done. Based on survey responses from marketing professionals, the model predicts the perceived success of core strategies such as Cold/Warm Calling, CRO, WhatsApp Marketing, and Online PR. The data were preprocessed following procedures including imputation for missing values, featSure scaling, and one-hot encoding for categorical and numerical variables. The multioutput neural network was trained for the regression task, achieving a test RMSE of 0.56, indicating high predictive power. To support the interpretability of the outputs, the continuous valued predictions were discretized into classes, so classification performance could be evaluated. The model achieved a superb overall classification accuracy of 97%, with respective F1-scores of 0.96, 0.93, 0.97, and 0.96 for Calling, CRO, WhatsApp, and Online PR. In addition, macro and weighted F1-scores held at 0.96, suggesting strong and even performance among all strategy types. Such results shed light on recognizing the worth of deep learning for marketing decision support based on data. Henceforth, the explainability of the model will be enhanced, concurrently with an exploration of classification-centric architectures with feature importance analysis.