Docs

The Iteration Engine

The Loop is what makes LarryLoop different. It's an AI feedback cycle that learns from your actual performance data — killing what doesn't work, doubling what does.

How It Works

Once you have at least 10 published posts with analytics data, the iteration engine kicks in. Every Sunday at midnight UTC it pulls your analytics, scores your posts, generates an iteration plan, and feeds winning and dead patterns back into the knowledge base for the next generation.

Pull analytics & score every post

Views (50%), engagement rate (30%), and download clicks (20%) are weighted into a score for each post.

Kill dead hooks

Hooks, styles, and topics that consistently underperform are flagged. They're never used again in future generations.

Double down on winners

High-scoring hooks and formats are extracted and used as the template for new variations in the next batch.

Experiment with new styles

Promising posts are explored further — small variations on themes that are working but haven't peaked yet.

Update the knowledge base

Winning and dead patterns are fed back in. The next generation uses this updated knowledge automatically.

Post Score Tiers

Every post is classified into one of four tiers based on its performance score:

DeadKill it

Low views, minimal engagement. This hook or style isn't resonating. The AI will never use similar patterns in future generations.

TestingWait and watch

Not enough data yet. Posts stay in Testing until they have enough views to evaluate properly.

PromisingExperiment

Decent performance, growing momentum. The AI tries variations on this theme to see if it can push it into Winner territory.

WinnerDouble down

High views, strong engagement, downloads. The AI extracts the hook style, visual format, and topic — and generates more like this.

When Does It Run?

Auto-iterate: Sunday midnight UTC
  • → Requires at least 10 published posts with analytics data
  • → Runs automatically for all paid subscribers
  • → Winning and dead hook patterns are saved to the knowledge base
  • → The next generation run (Sunday 2am UTC) uses the updated knowledge

The more posts you have, the better the iteration. Early weeks are data collection — the loop gets noticeably smarter after week 3–4.

What Gets Fed Back

The AI analyses patterns across all your posts, not just individual scores:

Hook style patterns

Which opening lines drive the most views — question hooks, shock stats, story openers, bold claims?

Visual styles

Which image style generates more engagement specifically in your niche?

Topic clusters

Which features or themes resonate most with your audience?

CTA effectiveness

Which calls-to-action actually lead to download clicks and revenue?

It Gets Smarter Over Time

Each iteration cycle the AI has more data to work with. After a few months of running:

  • Hook quality improves — the AI knows exactly which opening lines your audience responds to
  • Dead patterns are permanently avoided — no wasted posts on things that don't work
  • Winner variations are explored — small tweaks to proven formats to find new highs
  • Revenue signals are factored in — posts that convert get prioritized over posts that just get views

The bottom line: more data = better content. The longer you run LarryLoop, the more it compounds. A year in, you'll have a machine trained specifically on your audience.