Acceptable margin of error, response rate and sample representativeness

Concerning a survey in the context of social sciences, a questionnaire has been applied to the a sample of 477 companies (specialized in energy distribution in Spain, Universe is composed of 1129 companies). The criteria for selecting the sample is non probabilistic. The questionnaire has been sent to 639 communication managers and other representatives in this area, receiving 136 responses up to date.
In this context, which margin of error would be acceptable? What should be the optimum sample in order to be representative, and what should be the acceptable response rate?
Any recommendations of particular references or comments, would be welcome.