Sales professionals frequently face a difficult dilemma during outbound sales calls: whether to continue engaging a prospective client or end the conversation to pursue another lead. This screening decision is vital, particularly in settings where salespeople handle many leads but are constrained by limited time and frequent rejection. Despite its importance, research into how these decisions are made, their optimality, and ways to improve the process has been scarce. A recent study addresses this gap by examining decision-making in high-volume outbound sales, characterized by abundant leads, tight time limits, and high chances of call failure. The study applies an optimal stopping framework—a mathematical strategy that identifies the best moment to stop an activity to maximize reward. Researchers created a generative language model-based sequential decision agent, called a "stopping agent, " designed to learn when to end sales calls by mimicking an optimal stopping policy derived from past sales data. This AI tool analyzes complex textual data from calls, using natural language inputs to determine when to disengage. It supports integration with both open-source and proprietary large language models, highlighting its flexibility. Tested on outbound calls from a major European telecommunications firm, the stopping agent cut time spent on unsuccessful calls by 54%, a major efficiency improvement, while retaining nearly all successful sales despite shorter calls.

The saved time could be redirected to new calls, increasing expected sales volume by up to 37%. Further analysis revealed that salespeople tended to focus heavily on a few obvious linguistic signs of disinterest, overlooking subtler cues, which led to poor prediction of call failure risks and indicates cognitive biases limiting optimal real-time decision-making. These findings underscore the value of AI-driven tools like the stopping agent to overcome human cognitive constraints, enhancing time management and sales effectiveness. In today’s competitive market, optimizing sales contacts and agent time is critical, and AI systems provide a promising solution. Beyond sales, this research highlights AI's broader potential to improve rapid decision-making in complex, dynamic situations. Future studies might adapt this approach to other conversation-based contexts such as customer support, telemarketing, and negotiations, potentially boosting operational efficiency across industries. Overall, this study marks a significant advance in applying AI to optimize human decision-making in sales, enabling businesses to reduce wasted time and significantly elevate performance by intelligently deciding when to end unproductive interactions. The combination of human expertise and AI decision support promises transformative impacts on salesforce management and strategy.