When a video is published the TikTok algorithm needs to decide how much it should be shown and to who. To do this there needs to be a calculation of the video quality and an audience assessment classification.
A video will first be shown to a small number of viewers in between popular videos. This is often to a very targeted, engaged audience. Users that spend more time on the app will get shown more new videos whereas someone who’s only logged in for the first time and only stays on the app for 5 minutes is more likely to see a highlights reel.
This initial distribution is called the seed phase. While your video is seeding TikTok collects a number of data points.
- Average time spent on video and percentage of users that watched the full video.
- Video interaction:
- Profile interaction
- Clicks to your profile page
- Follows to your profile as a result of watching the video.
- Dropoff, percentage of users that left the app immediately or soon after watching your video.
These factors combined with account life-time metrics provide the video quality assessment which determines the reach your video will receive.
Content creators have described their TikTok videos as getting waves of traffic. The theory is that the algorithm works via a push-assess mechanism. An initial push of traffic is given to seed all videos, an assessment then takes place on that traffic and the best from each group gets pushed out as a 2nd wave. The metrics from the 2nd wave group of videos are assessed and the best videos get pushed out to a wider audience each time.
Another analogy that is often used is that of focus groups where TikTok is said to distribute your content to an initial group. Gain results and then if the metrics are good enough your content gets pushed out to a larger focus group for more accurate metrics across a wider audience.