This study investigates the effects of dynamic and static external human-machine interface (eHMI) designs on driver perception, usefulness, and satisfaction in autonomous vehicle (AV) scenarios, focusing on road bottleneck situations. A total of 30 participants were recruited to evaluate six eHMI configurations under varying dynamic states and AV action strategies. Results show that fully dynamic eHMI designs significantly improve information clarity, attention attraction, and interaction efficiency, achieving a correct interpretation rate of 72.68%, compared to 65.12% for static designs. Furthermore, AVs employing a “yielding” received higher ratings in usefulness and satisfaction, improving trust and decision accuracy. These findings underline the importance of dynamic eHMI in promoting traffic safety and interaction effectiveness, providing valuable insights for the development of future AV technologies.