Neural Deep Learning Model to Characterize the Brand Perception in Insurance Corporate Advertising

The evaluation of brand perception through corporate advertising is currently a challenge for companies since the effectiveness of traditional marketing methodologies depends on the ability of human interpretation, which can lead to un-objective perceptions. This paper proposes a methodology to characterize the brand perception for insurance companies based on a neuro-scientific methodology which uses a series of electroencephalographic signals (EEG – Emotive Epoc R) that gather the brain activity of a set of people (from 18 to 25 years) subjected corporate advertising. For analysis of brand perception or brand attributes, the methodology incorporates a Stacked Deep Learning model which has a Softmax function to classify the EEG signal according to four basic emotions. The internal layers of neurons that make up the model were configured using an auto-encoder learning strategy. The methodology reached accuracy indices close to 90% against the classification of EEG signals in four categories or basic emotions. These results allowed to characterize the brand attributes, establishing the general methodology to create novel travel insurance products for an insurance company based on emotions and taking as reference a scale of affinity according to the perceptions that an individual recognized in a set of corporate guidelines selected for this study.