Creator pattern trends
"What creator pattern is working this month, and has it changed?" — the monthly trend question that drives the next quarter's creative briefs.
The question, verbatim: "What pattern works in creators? And has it changed in last one month? … Last month AI came and disrupted. So that is the net new we saw last month. So is there any trend we spotted last month? Is generally the question."
This is the team's monthly pattern-spotting question. New trends in creator content (a new hook style, a new format, an AI-generated approach) get a few weeks of unfair lift before competitors catch on. Spotting them early is high-leverage.
Who asks
- Perf-marketing leader, in monthly planning
- Category / creative head, briefing the next month's creator slate
- Founder, when curious about the creative engine
Frequency
Monthly. Sometimes asked weekly, but the signal is only stable monthly.
Data you need
ad_creatives— every active and recently-paused creative- Performance per creative — spend, NRoAS, hook rate, hold rate, days live
- Creative metadata — format (carousel/video/static), length, hook type, talking-style, music presence, on-screen text density, etc.
- Creator metadata — size band (mega/micro/nano), creator persona, niche
- Trend tags — manually-assigned or auto-extracted (AI-narrated voiceover, before/after format, etc.)
Known gap: the team has not historically tagged creatives with trend metadata. Starting June onwards there's discipline around tagging via SegWise (auto-tag), but coverage is noisy. The platform will start with what's tagged and improve as tag quality climbs.
How to ask it
- "What's the top-performing creative pattern this month? Compare to last month and the trailing 3 months."
- "Which creative formats have the highest hook rate this month — and is any single format gaining share?"
- "What's the net-new creative trend that emerged in the last 30 days? Show me which creatives ride it."
- "For carousel vs video this month: which is winning, and is the gap widening?"
What you'll get back
A trend report, structured:
Top-performing patterns this month:
- AI-narrated before/after (video) — 14 creatives running this pattern; avg NRoAS 3.1 vs portfolio avg 1.7. New pattern — first appeared 3 weeks ago.
- Founder-led talking-head + product demo — 22 creatives; avg NRoAS 2.4. Stable for 3+ months.
- Micro-influencer testimonial carousel — 31 creatives; avg NRoAS 1.9. Slight decline (-12% vs prior month).
Pattern share-of-spend shifts:
| Pattern | Last month | This month | Δ |
|---|---|---|---|
| AI-narrated before/after | 2% | 11% | +9pp |
| Founder talking-head | 18% | 22% | +4pp |
| Micro-influencer carousel | 28% | 24% | -4pp |
| Mega-influencer reaction | 14% | 10% | -4pp |
Recommendations for next month's brief:
- Scale AI-narrated before/after — capacity 20-30 creatives
- Hold founder talking-head
- Reduce micro-influencer carousel briefs by 30%
The logic tree
Monthly creative trend pull
├── Identify candidate patterns (format × hook × creator-band combinations)
├── For each candidate:
│ ├── Cluster all creatives matching it this month
│ ├── Compute avg NRoAS, hook rate, hold rate
│ ├── Compare to portfolio avg
│ └── Compare to same pattern's prior-month numbers
├── Flag "net-new" patterns (first appeared <30d ago) separately
├── Surface declining patterns separately
└── Rank by expected next-month yieldHow to make it recurring
Save as a monthly report (not a tracker):
- Cadence: monthly, 1st working day
- Delivery: as a structured report into the perf channel + email to leadership
- Linked to: the monthly MBR
A weekly snapshot is too noisy — trend signals stabilise at monthly.
Pitfalls
- Tagging coverage gaps. If only 40% of creatives are tagged, the pattern recommendation is biased toward what's tagged. Spend a few hours fixing tag coverage before relying on this.
- Mistaking creator effect for pattern effect. A pattern might "win" only because a single high-performing creator happens to use it. The platform isolates this — but check the per-creator view if a recommendation feels off.
- Riding patterns past their decay. A new pattern's unfair lift lasts 4-8 weeks at most. Set a calendar reminder to revisit and rotate.
- Acting on n=3. A pattern with only 3 creatives isn't a pattern. The platform applies a minimum sample threshold; respect it.
Related
- Creator size mix
- Ad action decisions
- Product-lag RCA — when winners stop coming