This document presents a proposed approach for detecting suggestions in user reviews using a gated recurrent neural network (GRU) and convolutional neural network (CNN). The methodology uses a CNN to generate text embeddings, followed by a GRU and CNN layers to classify reviews as containing a suggestion or not. The approach is evaluated on a benchmark dataset, achieving an F1-score of 0.5806, outperforming other methods. Future work could leverage external knowledge and handle ambiguous/short reviews.