TL;DR
- Our pilot campaign ran for seven days in October 2026: 301 creators, 698 videos, 5,128,739 views. See the campaign report and measurement window.
- 113 of those videos passed 10,000 views. They are 16% of the videos and they brought 83.3% of the views.
- The 113 averaged about 37,800 views each. The other 585 averaged about 1,460.
- The campaign total hides a wide spread between individual posts.
In our pilot campaign 113 of 698 videos, about one in six, brought 83% of the views. The aggregate is real, but it does not show which post would perform well before publication.
16% of the videos brought 83% of the views
The campaign was a song release for a beverage brand. The brief asked for 600 to 700 videos from 200 creators in seven days, against a target of 4 to 5mn views. 301 creators posted 698 videos, 403 on Instagram and 295 on TikTok, and the campaign closed on 5,128,739 views.
Of the 698 videos, 113 passed 10,000 views. Those 113 account for 83.3% of all the views, about 4.27mn. Divide it out and each of them averaged about 37,800 views. The remaining 585 videos shared about 856,000 views, an average of about 1,460 each.
All the creators were briefed on one sound and posted in the same seven days.
The averages hide the variation between posts
The 113 posts above 10,000 views averaged about 37,800 views apiece. The other 585 averaged about 1,460. These are views per post, not a measure of creator compensation, production effort or the reason the posts performed differently.
The best prediction models explained less than half of how far posts went
The obvious answer is to hire only the creators whose videos will travel. We would like to. On this campaign we could not have told you which 113 it would be, and the research on the question is not encouraging.
Researchers including Duncan Watts studied how well success can be predicted in complex social systems, with an empirical test on the size of information cascades on Twitter. Despite an unprecedented volume of information about users, content and past performance, their best models explained less than half of the variance in cascade sizes. They conclude that even with unlimited data, predictive performance would be bounded well below deterministic accuracy.
The usual hedge is a bigger audience. A larger account may bring more viewers, but the next post’s reach is still uncertain. TikTok’s explanation of its recommendation system says follower count and previous high-performing videos are not direct ranking factors.
Read the full campaign report for the original brief, posting window and platform split.