Rebuffering (buffering, stalling) is when playback stops to refill the buffer because segments aren't arriving fast enough — the spinning loader mid-video. It is the single QoE problem most strongly tied to viewer abandonment and, for subscription services, to churn: viewers tolerate lower resolution far better than they tolerate a stall.
Rebuffering is usually measured two ways: the rebuffer ratio (time spent stalled as a share of total watch time) and rebuffers per hour (frequency). Both matter — one long stall and many short ones feel different. The causes span the whole stack: insufficient network throughput, an ABR algorithm that ramps too aggressively, CDN or origin problems, an encoding ladder with no low-enough rung, or a too-small player buffer.
Because it is so consequential, reducing rebuffering drives a lot of engineering: better ABR tuning, a well-designed ladder with a reliable bottom rung, multi-CDN steering away from degraded paths, and low-latency techniques balanced against the buffer headroom they sacrifice.
The handbook · 7 volumes
Fora Soft Video Engineering Handbook
Rebuffering is the stall viewers notice most, and it is the QoE metric most closely tied to abandonment and churn. We wrote seven volumes covering the video stack; for rebuffering start with Volume 3 on buffer-based ABR, CDNs, and playback recovery, then Volume 1 on bitrate ladders, with network transport in Volume 2. Written by Nikolay Sapunov, CEO at Fora Soft.
Rebuffering rates rising and viewers leaving mid-stream?
We will analyze your playback sessions and send a plan for player, ladder, and CDN changes with the expected drop in stalls. 250+ video products shipped since 2005.
