The interpretation of tendency surveys is very often based on changes in balances of opinion, which constitute the most commonly used summaries of the responses of companies. This article suggests the construction of different indicators, which, on the one hand, take into account "stable" or "normal" response conditions, and on the other hand also take into account the dynamics of individual responses from one survey to the next. The values of these indicators at the end of each month take into account not only companies' responses to the present survey but also their responses to the previous survey. The responses of companies during two consecutive surveys are classified into nine categories: Increase-Increase, Increase-Unchanged, Increase-Decrease, Unchanged-Increase, etc. We focus on the percentages of companies that change their responses from one survey to another. These percentages could give early signals of certain changes in the short-term economic situation and represent slightly advanced indicators. In particular, the article shows that, for certain questions, the percentages of Unchanged-Decrease and Unchanged-Increase responses do indeed display a number of tendencies earlier than the corresponding balances of opinion. It is, then, possible to construct an indicator based on these percentages, which could complete those which already exist for forecasting the quarterly growth rate of manufacturing production over two quarters.