Companies make use of the voice of the customer in their product development work and their efforts to increase customer satisfaction. To use the voice of the customer more effectively, companies need to be able to classify text content representing the voice of the customer and direct it toward the pertinent department. Various methodologies have been proposed, but classification errors occur because of ambiguity in the meaning or interpretation of text. In the present research, application of parameter design to the classification of the voice of the customer improved the robustness of classification performance with respect to textual variations and reduced the number of man-hours spent on classification by a factor of four. In addition, when the results were tested by the F-measure, which is widely used in text classification, a high S/N ratio was found to be associated with a high F-measure, confirming the usefulness of the S/N ratio in the field of text classification