How Klypps measures

Every screen, export and answer uses the same scoring engine.

1. Usual: the median views of the brand’s own posts on the same account and format (Reels, Shorts, long videos, other videos) in the 90 days before the post, excluding the post itself. Organic, paid and creator posts are kept apart.
2. A median keeps one viral post from distorting normal performance. Sponsored, paid and creator posts stay separate.
3. × usual: a post’s views divided by its usual, shown to two decimals. With fewer than 8 earlier posts the figure is shown lighter and marked “early”. Images, carousels and ads are not scored on views.
4. Illustration: 1.80× usual means 80% above normal; 1.00× means exactly normal.
5. Category score: a post compared with all tracked brands on that platform over the prior 90 days. Needs 10+ posts from 2+ brands.
6. Confidence: Proven (30+ posts), Early (10–29), Too few (under 10).
7. n is always shown.
8. Hit: a backed piece that beats its usual (above 1.00×) within 14 days of posting.
9. Hit rate: hits ÷ measured backed pieces, shown against the baseline. Under 10 backed pieces it is marked early. 7- and 28-day results are checkpoints, not the hit rule.
10. What AI reads in a video: hook and strength, story beats and payoff, first logo and product appearance, product screen time, cuts per second, language mix, format, call to action, subtitles, trending audio where verified, creator on screen, emotional curve and peak, primary emotion and tone. Comments supply sentiment, emotion and themes.
11. Public data only until accounts are connected. Views refresh daily, not live. Patterns show what tends to beat usual, not proof of cause: read n and the label. AI tags can be wrong; analysis carries a version and confidence. Claims without supporting posts are blocked.