In this article
After reading, you'll understand:
- What buffering is and why it happens
- Initial buffering vs rebuffering, and the metrics for each
- How adaptive bitrate and a CDN prevent it
- What you can (and can't) do to eliminate it
What is buffering?
Buffering is when video playback pauses to download more data before it can continue. The player keeps a small reserve of upcoming video (the "buffer"); when data arrives slower than it's played, that reserve empties and playback stalls — the spinning wheel — until enough has loaded to resume.
It usually comes down to one of a few causes: the bitrate is too high for the viewer's connection, the server or CDN is far away or overloaded, or the video wasn't prepared for adaptive delivery. Rebuffering is one of the strongest predictors of viewers abandoning a video, which is why so much of streaming's technology exists to avoid it.
Initial buffering vs rebuffering
Not all buffering is equal, and the industry measures the two kinds separately as quality-of-experience metrics.
| Type | When it happens | Metric | Impact |
|---|---|---|---|
| Initial buffering | Before playback starts | Video start time | Viewers abandon before watching |
| Rebuffering | Mid-playback stalls | Rebuffer ratio (% of watch time stalled) | Most damaging — interrupts committed viewers |
A low rebuffer ratio and a fast start time are the two numbers that best predict watch-through — worth tracking in your analytics.
How buffering is prevented
The player steps quality down before the buffer empties, so playback continues at a lower resolution instead of stalling — then steps back up when the connection recovers.
Serving segments from an edge server near the viewer keeps data arriving quickly and reliably, keeping the buffer full even under load.
A well-built rendition ladder gives the player realistic rungs to pick from, so it can always find one the connection can sustain.
Holding a few seconds of video in reserve lets the player ride out short bandwidth dips without stalling — a trade against latency on live.
How Kinescope handles this
Kinescope is built to keep rebuffering close to zero: every upload is transcoded into an adaptive ladder, delivered from Kinescope's own global CDN, and played through an adaptive player that steps quality to match each viewer's bandwidth. Per-viewer analytics surface start time and drop-off, so you can see where playback problems cost you audience — no configuration required.
Read the Kinescope docsFrequently asked questions
Why does my video keep buffering?
Buffering happens when the player can't download data fast enough to keep playing — usually because the bitrate is too high for the connection, the server (or CDN) is far away or overloaded, or the video wasn't prepared for adaptive streaming. Adaptive bitrate fixes most of it by stepping quality down before the buffer empties.
What is the difference between buffering and latency?
Buffering is playback pausing to load more data mid-stream. Latency is the delay between something happening (or you pressing play) and seeing it. A live stream can have low latency but still buffer on a weak connection, and vice versa — they are different problems.
What is the difference between initial buffering and rebuffering?
Initial buffering is the wait before playback starts (measured as video start time). Rebuffering is a stall in the middle of playback after it has already begun. Both hurt, but rebuffering is the more damaging: a viewer who has committed to watching and then gets interrupted is far more likely to abandon.
How does adaptive bitrate prevent buffering?
Adaptive streaming encodes the video at several quality levels and lets the player drop to a lower one when bandwidth falls, so playback continues smoothly at reduced quality instead of stalling. It steps quality down before the buffer runs dry, then back up when the connection recovers.
Does a CDN reduce buffering?
Yes. A CDN serves video from an edge server near the viewer, so segments arrive faster and more reliably — which keeps the buffer full and reduces stalls, especially under load or for distant viewers. It's one of the biggest levers on rebuffering at scale.
Can I eliminate buffering completely?
Not entirely — a viewer with almost no connection will eventually stall. But you can make it rare: adaptive bitrate, a fast CDN, sensible bitrate targets and a healthy buffer window together keep rebuffering close to zero for the vast majority of viewers.
Further reading
- Adaptive bitrate streaming — Wikipedia — the main defence against buffering.
