Startup score
A startup score grades how quickly a video began playing after the viewer pressed play. Short waits score close to 100 and the score falls as the wait grows. It turns a raw startup time in milliseconds into a number that can be compared across titles, devices and days.
Why it matters
Startup is the first impression of every view. Slow starts lead viewers to give up before the video plays, which means lost views, lost ad impressions and lower engagement. A score makes startup easy to track against a target and to compare across devices, titles and app releases.
How it works
- The player notes the moment the viewer presses play or autoplay begins.
- It notes when the first frame of video actually appears on screen.
- The gap between the two is the startup time, measured in milliseconds.
- A formula converts that time into a score, so short waits score near the top and long waits fall steadily.
What is a good video startup time?
A good video startup time is under 2 seconds, and many short video and news apps aim for 1 second or less. Research on large video audiences found viewers begin to give up once a start takes longer than about 2 seconds. Live streams and DRM-protected titles often take a little longer, so set separate targets for each.
How do I reduce video startup time?
You reduce video startup time by making the first seconds of video arrive and decode faster. Start on a lower rendition and switch up once playing, use shorter segments, keep the CDN close to viewers, preload the player on pages where play is likely, and trim heavy scripts and plugins around the player. Measure each change with real viewer data. Each change can be measured against real viewers in FastPix Video Data, since a release that slows startup shows up as a drop in the startup score.
Where it's used
Short video apps and news publishers track the startup score closely, because viewers scrolling a feed give a clip only a second or two before they move on.
Where it usually goes wrong, and how FastPix handles it
Raw startup times are skewed by a few very slow views, so averages swing widely and teams struggle to set a clear target.
The formula in FastPix Video Data is 8 / (8 + startup seconds), so a 2 second start scores 0.8 and an 8 second start scores 0.5. The startup score carries a 0.275 weight in the QoE score. That gives product teams a steady target, and a slow release shows up as a clear drop instead of noise.
By the numbers
- In a 2012 study of 23 million video views, viewers began abandoning videos that did not start within 2 seconds, and each extra second of delay raised abandonment by about 5.8 percent. (Research quantifies how video stream quality affects viewer behavior (ScienceDaily, 2012))
