Extreme voices are frequently spotted to have a significant effect on public opinions on online forums or social media, but very little is known about when, what, and how to tone it towards softened and how to manage them. To deal with the large volume of extreme opinions, companies, and PR authorities spent heavily on soothing the tension and sheltering customer word-of-mouth from devasting reviews by feeding public overloaded information, dominating public attention and grabbing new headlines, yet very few studies rationalize this assumption. This paper develops simulation experiments to unmask the aforementioned relations, in particular, how information environment (i.e., information preference and noise), individual factors (i.e., acceptance threshold, halo effect), and institutional efforts (i.e., promotion) affect the customer review’s evolution at the aggregate level. The findings suggest that the evolution of opinion in divergence can occur in any preference-held audience whereas the information noise can substantially moderate the information valence and cause the polarized voices to get mild dramatically. In addition, a halo effect on the information source can stir the voice to be amplified by the influencer. The experimental results indicate the significance of understanding information valence and group conformity in conversational contexts and find a few clues of improving communication effects. This research bears high originality in explaining how public opinions evolve in a belief-neutral environment via audience-sensitive, context-bound, objective-manipulated techniques. The findings suggest valuable meanings to review monitor and control in information management.