State the simple geometry at the core of it. A flat ground antenna pointed straight up sees a cone of sky overhead, but low-Earth-orbit satellites do not spend their time evenly across that cone, the favorable geometry depends on the terminal's latitude and the constellation's orbits. Tilting the antenna biases its field of view toward where the satellites actually are.
SpaceX's grant US12244399B2 (inventors including Duncan E. Adams and David Milroy), classified across H04B 7/18519, H04B 7/18517, and H04B 7/18547 (satellite-communications acquisition and link management), claims tilted earth-based antennas and the methods of tilting them. The acquisition-and-link CPC cluster shows it is a patent about maximizing useful sky, not about the antenna's internal RF design.
The mechanism pays off in link time and quality. By orienting the terminal so its best-performing region of coverage overlaps the satellites' most-traveled paths, the antenna keeps a usable link for more of each pass and sees more satellites to hand off between, all without adding hardware, just smarter geometry.
It extends the user-edge story SpaceX has been telling in patents for years: make the terminal cheap (uni-dimensional steering), help it find satellites (the 2024 visibility grant), and now orient it to see the most productive sky. Three documents, one strategy, wring maximum performance from an inexpensive, mass-produced dish through clever system design rather than costly hardware.
The honest limit is that the optimal tilt depends on location and constellation, and a fixed tilt is a compromise across conditions. The patent tells you SpaceX treated antenna orientation as a tunable performance lever worth protecting. The realized benefit depends on where the terminal is installed and how the constellation overhead is arranged.