Playbook replay
Winning-format benchmarking
Reads the top posts in a category and returns the hooks, lengths, and formats that actually win there.
Run for Lumira, a demo brand we invented for this page. No customer data. Nothing here runs against a live account.
Every session here was run once and saved, against a demo brand or public data. On your own account Naise works from your brand voice, your campaigns and your connected channels, so it reasons about your data instead of a stand-in.
Ranking by raw views would just surface the biggest accounts. I should score each post against its own author's median so I am measuring the format, not the follower count.
Pulling the top posts on the tag, then reading them as a set rather than one at a time. Patterns only show up across the whole sample.
I scored each post against its own author's median views, so what follows is about the format rather than the account size.
Length
| Duration | Share of top posts | Median vs author baseline |
|---|---|---|
| Under 15s | 18% | 1.1x |
| 15 to 30s | 54% | 2.3x |
| 30 to 60s | 24% | 1.4x |
| Over 60s | 4% | 0.8x |
The 15 to 30 second band is not close. Posts over a minute underperform their own author's baseline in this niche.
Hook shape
Three constructions account for most of the outperformance:
- Problem stated in the first two seconds, on camera, no intro. 2.8x baseline.
- Contradiction of a common belief, for example "stop double cleansing". 2.4x baseline.
- Number promise, for example "three steps, four minutes". 1.9x baseline.
What does not work here: brand name in the first line, and any opening that starts with a greeting.
What to model, concretely
- Shoot 15 to 30 seconds. If a script runs to 45, cut the setup, not the payoff.
- Open on the problem, in frame, no logo card. The first frame with a logo costs about a third of the watch-through in this sample.
- Text on screen in the first second. 87% of the top posts had it, and this niche is watched muted.
- One step per post. Full routines underperform single-step posts by a wide margin, which is counterintuitive for a tag literally called routine.
What not to copy
Two of the highest-view posts in the sample are outliers driven by a trending sound, not by format. They are worth watching but not worth modelling: the sound is already past its peak, and building a plan on it means arriving late.
Naise created an artifact
The hooks, lengths, and formats that beat their own author's baseline, with posts to model.
Model these
| Pattern | Lift vs baseline | Example |
|---|---|---|
| Problem in first 2s, on camera | 2.8x | "My skin barrier was gone." |
| Belief contradiction | 2.4x | "Stop double cleansing." |
| Number promise | 1.9x | "3 steps, 4 minutes." |
Avoid
- Logo card in frame one
- Greeting openers
- Full routines in one post
- Over 60 seconds
That was a sample. Run it on your own brand.
Naise reads your brand, your channels, and your market, then does this against your real data.
Run this playbook free