Experimental protocols testing the effectiveness of cheap talk are numerous but have generated conflicting results. The theoretical interpretation of hypothetical bias as a strategic response according to the perceived consequence could be the missing key to understand these opposite results from the literature. Increasing evidence suggests that this bias rises from subjects’ perception of how stated preferences surveys will be used; some subjects believing that stated valuations can impact the price of the good, while others that it will influence its provision. Subjects strategically respond by adjusting their declared values accordingly. This paper reports experimental findings supporting the presence of strategic response, showing that cheap talk operates by mitigating these behaviors and potentially explaining cheap talk's heterogeneous results..