# Retention Studio strategy for mechanism-explainer Shorts

Use this when a shorts project should move beyond generic AI-video output into a semi-automatic retention studio.

## Core distinction

Do not optimize for "KI-Look". Optimize for retention mechanics:

- first 1-2 seconds open a loop
- visual change every 1-2 seconds
- every scene reveals an invisible mechanism
- the viewer ends with: "I want to see more of these systems"

## Recommended positioning

Best fit for this user's projects:

```text
Hidden Money & AI Systems
Series: Unsichtbare Systeme in 30 Sekunden
Claim: Wir zeigen Dinge, die dich Geld, Zeit oder Kontrolle kosten – bevor du sie bemerkst.
```

Avoid pure KI-avatar news, static AI slideshows, reused-content clones, or generic "what if" videos.

## Mechanism Explainer formula

```text
0.0-1.5s  Pattern interrupt
1.5-5s    Problem
5-18s     Mechanism reveal
18-28s    Twist
28-35s    Follow trigger / series promise
```

The key visual mechanic is: **show the invisible mechanism**.

Examples:

- car price -> x-ray layers: repair risk, stand days, margin, financing
- AI agent -> control loop: goal, tools, approval gate, runaway path
- building project -> timeline: permits, objections, procurement, inflation, scope creep
- bad deal -> brain/courtroom: sunk cost, confirmation bias, loss aversion

## Idea scoring dimensions

Score ideas for:

- surprise
- data basis
- comment potential
- series potential
- monetization potential
- production effort penalty
- specific visual mechanism bonus

Produce only high-scoring ideas. Develop medium ideas. Reject generic montage ideas.

## Production principle

Build artifacts in this order:

```text
strategy -> scored ideas -> mechanism explainer -> mechanism shot plan -> render manifest -> asset plan -> final renderer
```

Mechanism explainers should be adapted into production shot plans without losing beat timing, visual motion, sound cues, or the no-static-cards rule. The shot plan is where `pattern_interrupt`, `problem`, `mechanism_reveal`, `twist`, and `follow_trigger` become renderer-facing scenes. When converting those shot plans to render manifests, preserve `visual_motion` and `sound_cue` on timeline items; these are not decorative fields, they are the retention/editing contract for the final renderer.

For final render, use motion-led asset modes:

- cinematic AI video
- screen-recording simulation
- animated card
- kinetic text overlay

Keep static cards at `0` for final output.

## Testing checks

Tests should assert:

- strategy is not positioned as AI-video factory
- `static_card_limit == 0`
- mechanism explainer has all five beats
- pattern interrupt ends within 1.5 seconds
- every beat has visual motion and sound cue
- weak generic ideas are rejected
- pilot ideas cover money, AI, Swiss systems, and thinking-error pillars