Average bitrate played is the time-weighted mean of the quality rungs viewers actually received during a session, computed over playing time only so that stalls do not count. It is derived from the player's logged rendition changes rather than from a standard media event, and it signals how effective the ABR ladder and adaptation logic were at delivering picture detail across the session. It is the steadiness metric's counterpart to bitrate switching: a higher average usually looks better and matters more for live than for on-demand content, but it is an effectiveness signal, not a picture-fidelity score, so higher is not automatically better-looking for a given piece of content, and it says nothing about whether the viewer saw it without stalling. Its catch is that it must always be read against the rebuffering ratio and switching, because the same ABR choice that lifts the average bitrate can also drain the buffer and invite a stall, the trade-off ABR makes every segment.
