This thesis introduces the concept of \textit{strategic differentiation} to explain how authoritarian regimes, rather than issuing uniform propaganda, adapt their messaging across platforms and audiences in a deliberate, calibrated manner. Drawing on Chinese state media texts from 2007 to 2022—including China Central Television (CCTV), People’s Daily, and Government Work Reports—this study examines how language, framing, and interpretive accessibility vary depending on audience sophistication and political context. Using semantic network analysis with FastText word embeddings and entropy-based readability metrics, the study finds systematic evidence of discursive segmentation. Elite-facing platforms employ lexically rich, technical language, while mass-facing outlets rely on simplified, emotionally resonant narratives. Temporal analysis reveals that propaganda coherence intensifies during crises and relaxes during routine periods, suggesting that message alignment is a strategic choice rather than a constant feature. These findings contribute to the literature on authoritarian information control by highlighting how regimes manage interpretation—not only through content, but through audience-specific messaging strategies. While the study is limited in its focus on textual media and does not directly measure audience reception, it provides a structured framework and replicable method for analyzing differentiated propaganda. The concept of strategic differentiation offers a new lens for understanding how authoritarian regimes communicate selectively to maintain control, manage legitimacy, and contain dissent.