Iterative content adaptation systems attempt to conserve bandwidth required to download multimedia content on mobile devices by initially loading an adapted version of the original content and refining its quality only when deemed necessary and explicitly requested by the user. In this paper, we demonstrate that iterative approaches that only minimize bandwidth consumption may result in latency for browsing adapted web content that are even higher than when no adaptation is performed. We present PAS, an adaptive algorithm that can optimize browsing latency by taking into account both expected download time and user interaction time. Our experiment shows that PAS reduces browsing latency by up to 65% and reduces bandwidth consumption by up to 80%. Furthermore, PAS is also flexible and can accommodate user preferences for trading off bandwidth consumption and user interaction time.