In short
An outlier outperforms comparable videos from its own account, rather than simply having many views. Use Only outliers to identify these references, then examine their content. Without enough history, a missing badge does not mean poor performance.
Read the signal
The threshold is at least twice the reference median, using at least 10 comparable posts from the same account, network and source type. Video age and duration must stay within a 20% range.
This signal compares a post with its account. It does not compare every brand in a niche.
The median represents the middle of the reference sample: it prevents one exceptional video from defining the usual level by itself. The target video is excluded from its own references. Comparing similarly aged videos also avoids artificially favoring older posts.
Show outliers
- In Discover, start with a relevant search. Enable Only outliers in filters.
- Open a matching video and examine its product, hook and available metrics before saving it.
First compare the list with and without Only outliers while keeping the same search and other filters. Expect a shorter selection of qualifying posts, not a new collection or a rating for every excluded video.
Example of reading a multiple
Fictional example: at a comparable age, a video has 30,000 views and its reference median is 10,000. Its ratio is 3×; it exceeds the 2× threshold if the other comparability conditions hold. Another video with 100,000 views on an account whose median is 150,000 does not outperform by this signal.
After this initial screening, watch the content: which action, hook or demonstration deserves study? The multiple helps you choose a reference; it does not establish which element caused the increase.
When data is missing
Insufficient history prevents qualification. A post without a badge should not be treated as a failure. An outlier proves neither advertising profitability nor sales; use it as a reference to investigate.