After the arrival of social networks and the interconnection that it caused among its users, the generation of businesses within these platforms was inevitable, the main one of them selling advertising, in Ecuador the sale of advertising through social networks amounts to a amount exceeding thirty million dollars only for companies in the commerce sector, therefore, this paper analyzes the efficiency of advertising spending through the condition of Dorfman-Steiner for eighty-eight companies in the commerce sector during 2018, generating indicators of Management for digital marketing (KPI’s) using Machine Learning techniques for the processing of data from the social network twitter and relating them to the financial results of these companies during the same period, through a multiple linear regression. In the analysis performed, a significant effect was found by the indicators towards the Dorfman-Steiner condition for companies with a small number of tweets, the greatest effects found were given through the interaction between the indicators concluding that, to reduce the level Advertising spending should be aimed at the propagation and popularity of the content that is published, taking into account the quality of the content that is disclosed.Key Words:Dorfman-Steiner, KPI’s, machine learning, twitter.