What 780,000 UGC videos taught us about consumer apps
Three patterns that show up across categories — productivity, sleep, finance, AI tools — and what they mean for the next wave of UGC strategy.
Over the past 18 months we’ve analyzed roughly 780,000 TikTok and Instagram videos across about 1,300 consumer apps. The dataset spans verticals we never originally planned for — meditation apps, expense trackers, AI photo editors, sleep coaches — but the recurring patterns are surprisingly portable.
Three of them keep showing up. Worth a longer write-up on each later; for now, the short version.
1. Hook variance dominates view variance
Within a single account, the difference between a top-quartile and a bottom-quartile video is mostly explained by the first two seconds. Not the topic, not the creator, not the music. The hook.
We tagged a sample of about 15,000 videos with a structured hook taxonomy (visual hook + text/audio hook). Apps that ran 3+ distinct hook patterns at scale had a meaningfully higher median view count than apps that stuck to one pattern. The takeaway isn’t “test more hooks” — that’s stale advice. The takeaway is that hook diversity is a leading indicator of a healthy UGC program. If a brand’s last 30 videos all open the same way, the program is fragile.
2. Format and trend are almost orthogonal
Two of the dimensions we tag are format (Talking Person, Tutorial, Voiceover, Day-in-Life, etc.) and trend (recurring narrative arcs like “9-to-5 Rant” or “Reset Routine”).
A naïve guess would be that some formats correlate with some trends. They don’t — at least not strongly. The same trend gets ridden across every format. The same format hosts every trend. Practically, this means format and trend should be picked independently in any creative brief; don’t anchor one on the other.
The exception is Carousels, which have a tighter format-trend coupling because the medium itself constrains the story shape. We’ll write that up separately.
3. The viral threshold isn’t a number — it’s a slope
Most teams treat “viral” as a fixed view count: 1M, 5M, etc. The data doesn’t agree. What actually matters is the velocity — how quickly views accumulate in the first 48 hours.
A video that hits 1M views in three weeks has very different downstream characteristics (search lift, install conversion, creator follow-on) than one that hits the same number in 36 hours. The viral threshold worth tracking is velocity past a 5K/hour slope, not absolute counts.
Apps that build their UGC scorecards around velocity (not totals) make better creative bets. Three of the four fastest-growing apps in our dataset run that scorecard.
We’ll go deeper on each of these in future posts. If you want the methodology — how we sample, what’s in the tag taxonomy, how we handle confounders — let us know which one matters most. We’ll write that one first.