Researching a Niche with `last30days`¶

The fastest way to find out what's currently working in a niche: ask Claude to research it via the last30days skill. 30-minute output that would take you a day of scrolling.

What you get from a niche research pass¶

Across Reddit, X, YouTube, TikTok, Instagram, Hacker News, Polymarket, and web search (last 30 days only):

Top hooks observed with frequency and engagement signals
Recurring formats — talking head / mixed-media / POV / etc.
Trending angles — emotional or narrative entry points working right now
Specific creators worth modeling with follower count and a representative video
Copy-paste-ready prompts to hand directly to the Script Writer

How to invoke¶

You: research what hooks are working for magnesium supplements in
     the last 30 days. Focus on TikTok and Instagram for women 35-55.

Claude: [triggers last30days, scans listed sources, synthesizes findings]

Trigger phrases the skill recognizes:

"research [topic] from the last 30 days"
"last30 [topic]"
"what's trending in [niche]"
"what are creators doing for [topic] right now"

What a brief should specify¶

The more specific you are, the more targeted the output:

Niche / topic: "magnesium supplements", "perimenopause sleep loss", "AI avatar UGC"
Audience: "women 35-55", "Gen Z men with anxiety", "professional moms"
Platforms (optional): "TikTok and Instagram", "just TikTok", "include Reddit"
Goal (optional): "find new hook formats", "understand competitive landscape", "find creators to model"

A specific brief produces a specific report. A vague brief ("research wellness content") gets you a vague summary.

Reading the output¶

A typical research report has sections roughly like:

TOP HOOKS OBSERVED:
1. Authority-betrayal opener: "My doctor never told me this..."
   - Used by 7+ creators in the last 30 days
   - Average 80K views per post; one hit 2.1M
   - Maps to Solution Aware on the Schwartz scale

2. Symptom-cluster cold open with on-screen text
   - Lists 3-4 symptoms before the avatar speaks
   - Heavy use of bold serif captions, animated word-by-word
   - Average 60K views; one hit 1.2M

3. Confessional time-travel: "10 years ago I had no idea..."
   - Personal vulnerability opener
   - Mid-pack engagement but high save rate

RECURRING FORMATS:
- 80% talking head with caption overlays
- 15% mixed-media (PiP recipe demos)
- 5% pure B-roll with voiceover

TRENDING ANGLES:
- Perimenopause sleep loss (rising)
- Hormone-friendly bedtime routines (rising)
- Cortisol-disconnect angle (declining — used heavily 60 days ago, now cooling)

CREATORS WORTH MODELING:
- @creator-1 (327K followers, 80K avg views, uses authority-betrayal)
- @creator-2 (200K followers, 50K avg, mixed-media specialist)
- @creator-3 (145K followers, 100K avg, time-travel hook specialist)

COPY-PASTE PROMPTS FOR THE SCRIPT WRITER:
[3-5 specific prompts ready to feed into Stage 2]

What to do with it¶

The output is input for downstream stages, not just reading material:

Pick 1-2 hook formats to test (don't try to run 5 different formats at once)
Pick 1-3 creators to study deeper via creator-scanner
Save the research to Assets/{workflow}/research.md so the brief stage has it
Pass the copy-paste prompts to the Script Writer when scripting begins

When the output is useful but generic¶

Sometimes last30days produces a report that's accurate but not specific enough. If that happens:

Re-run with a tighter brief — narrow the audience, the platform, the goal
Cross-reference with creator-scanner on the specific accounts the report names
Don't act on generic findings — they'll produce generic scripts

A common pitfall: research says "hooks about menopause are working." That's a topic, not a hook structure. Drill down: which specific hook structures are working. Then act on those.

When research isn't worth doing¶

Skip research when:

The workflow is a straight variant of a winning workflow (you already have a working template)
The workflow is a video copy of a specific source (you're modeling that source, not the broader niche)
The brief is highly constrained (specific framework, specific format the brand has already validated)

Research is most valuable for new niches, new accounts, and format pivots where you don't have a recent winning template to lean on.

Combining with the Schwartz framework¶

A research report becomes more powerful when you overlay the Schwartz framework:

The report says "authority-betrayal hooks are dominant." That's a Solution Aware hook structure (audience knows about solutions but is being told the conventional advice is wrong).

If your account is targeting Unaware women in their early 40s (most haven't connected sleep loss to perimenopause yet), you have an opportunity — the dominant format may not be reaching the audience you're going after. A Problem Aware hook ("if your sleep has changed in your 40s, here's what's actually happening") might cut through where authority-betrayal can't.

See Schwartz & Why Research for the framework.

Frequency¶

How often to run research depends on the niche:

Fast-moving niches (AI tools, viral fitness, trending topics) — weekly
Steady niches (wellness, supplements) — biweekly to monthly
Slow niches (B2B SaaS, professional services) — quarterly

Don't run it for every workflow — that's overkill and burns time. Run it when:

Starting a new niche
Starting a new account in a niche
Noticing your existing workflows are stale (declining engagement)
Before a major variant push

When you're ready¶

→ Next: Scanning a Creator — deeper dive on a specific creator the research surfaced, to understand their format in detail.

Researching a Niche with last30days¶