Proactive behavior, dialogue personalization, and backchannels are essential elements for enhancing rapport, fostering engagement, preventing conversation stalls, and supporting a natural dialogue flow. This study integrates these behaviors into a rapport-building dialogue strategy using the CO-STAR and few-shot frameworks to prompt large language models (LLMs) within a human-robot interaction. The success of the integration was assessed through metrics such as conversation stall frequency, dialogue similarity, and backchannel usage by the robot. We then observed the effect of integrating these elements on participant-observed behavioral outcomes, including participant backchannel usage, self-disclosure, and subjective questionnaire responses. Experimental results indicated that incorporating these behaviors into the dialogue strategy positively impacts human-robot rapport both behaviorally and subjectively.