Driven by algorithm-driven platforms reshaping cultural consumption, book marketing has been fundamentally transformed by the integration of content recommendation, social interaction, and transaction conversion. After Douyin, a Chinese short-video platform, incorporated Gross Merchandise Value per Mille (GPM) into algorithmic weighting in 2024, the conventional linear model of "editing, printing, and distribution" has proven incompatible with instantaneous platform feedback. Against this background, the present paper proposes "marketing pre-positioning" and empirically investigates the Zhejiang Xinhua Bookstore Group's "Thousand Hosts" project, a pioneering digital transformation initiative within China's traditional publishing distribution system. This study deconstructs the triple predicament facing conventional publishing marketing through diagnostic analysis and case observation. First, traffic volatility and algorithmic cut-off mechanisms disrupt linear release cycles, disrupting the alignment between promotional promises and reading experiences. Second, elite-oriented discourse fails in interactive environments, diluting brand identity and engagement. Third, production structures centered on professional judgment cannot respond to fragmented audiovisual demands, compressing market lifecycles for content lacking visual translation. On this foundation, the research constructs a "marketing pre-positioning" framework with three interlocking dimensions. The first establishes "market-data-pre-positioning" by deploying AI agents to analyze competitor traffic, reader profiles, and trending cycles during acquisition pre-planning, transforming subjective research into evidence-based decision-making. The second implements "dual-track parallel" content translation grounded in IMC and SOR theories. The third constructs "public-private domain synergy" by integrating broadcasting modes and KOL/KOC alliances for public-domain distribution while channeling traffic to the Qiantang Hongshu App for private-domain accumulation. The study documents three core pathways. First, narrative translation reduces communication thresholds by transforming textual value into visual symbols, exemplified by the Cangqiong Shan Ji campaign achieving over 5 million exposures through a viral "cucumber doll." Second, matrix-based competition across nine vertical categories realizes scaled distribution, as demonstrated by the Jian Lai series achieving 3.7 million yuan and Aobing Zhuan selling 50,000 copies for 2.8 million yuan. Third, data-intelligent feedback loops feed user behaviors from livestream rooms back into selection decisions, shifting marketing from experience-driven promotion toward evidence-based management. Furthermore, immersive livestreaming demonstrates that cultural-field construction is crucial for high-involvement products. The Harry Potter 25th Anniversary Edition exceeded 2 million yuan in single-session GMV with a 5% return rate, substantially below the 20%–30% industry average return rate. These results confirm that traffic acquisition and transaction conversion can be integrated through organizational reconstruction. The findings indicate that "marketing pre-positioning" has significantly elevated core GPM by embedding transactional-efficiency indicators upstream into planning, restructuring the production-market feedback connection and driving a shift from passive distribution to proactive control. Accordingly, this research argues that publishing institutions can build algorithmic ecosystem advantages through talent activation, organizational restructuring, and AI empowerment as a full-domain paradigm, providing a replicable model for industry competitiveness and cultural dissemination in the algorithmic era.