Modo
Agendar llamada

trial-reel-distribution: distribuir tus reels ganadores

Usa los reels de prueba de Instagram para llevar tus reels que ya funcionaron a gente que no te sigue, sin editarlos, y pasa al feed los que superan tu vara de alcance.

trial-reel-distribution: distribuir tus reels ganadores

Respuesta rápida

Qué es
trial-reel-distribution es una skill de Single Grain (licencia MIT) que le enseña a Claude a usar los reels de prueba de Instagram como canal de distribución: reposta tus reels ganadores sin editarlos y pasa al feed los que superan tu vara.
Qué le das
Tus reels de los últimos 90 días ordenados por alcance, los archivos originales y una vara de alcance para decidir ganadores
Qué te devuelve
El banco de reels a reutilizar, el calendario de pruebas (unas 5 por día), el registro de alcance y la lista de ganadores que vuelven al feed
Licencia
MIT
En GitHub
3,8 mil estrellas · actualizada el
Revisada por Claura

Instálala en una línea para todos tus proyectos, en Claude Code, Cursor, Codex y más:

npx skills add ericosiu/ai-marketing-skills --skill trial-reel-distribution -g

Funciona en

Claude CodeEn ~/.claude/skills/
CursorEn ~/.agents/skills/
CodexEn ~/.agents/skills/

Pedidos para probarla

Con la skill instalada, copia uno y pégaselo a tu agente tal cual, o cambia los datos por los tuyos.

  • Ordena mis reels de los últimos 90 días por alcance y dime cuáles volver a subir como reels de prueba esta semana.

  • Define una vara de alcance para mi cuenta de 8.000 seguidores y dime cuándo pasar un reel de prueba al feed.

  • Arma el calendario de reels de prueba para el próximo mes con mis 10 reels que mejor funcionaron.

#Qué hace

trial-reel-distribution es una skill de Single Grain, la agencia de Eric Siu (licencia MIT), que le enseña a Claude a usar los reels de prueba de Instagram (los trial reels, que se muestran solo a gente que no te sigue) como un canal de distribución y no como un laboratorio.

Su razonamiento es concreto: un espacio de prueba es alcance frente a desconocidos, y gastarlo en un video que nunca funcionó casi siempre confirma, ante un público más frío, que no viaja. En cambio, tu feed ya hizo la prueba: los reels con más alcance de los últimos 90 días son contenido probado. La skill los vuelve a poner frente a gente nueva.

El método tiene siete pasos:

  1. Armar el banco: ordenar los reels del feed por alcance de los últimos 90 días, tomar unos 30, sacar los que ya se usaron y anotar el alcance original de cada uno.
  2. Preparar cada reel: el video queda exactamente igual (sin cambiar velocidad, música, gancho ni cortes), pero se exporta de nuevo para que no sea un archivo idéntico, porque los duplicados exactos se rechazan. El caption se escribe de nuevo, mismo tema y mismas reglas de pedido.
  3. Programar en tandas: unas 5 pruebas por día, repartidas, empezando por los de mayor alcance.
  4. Leer resultados: alcance a las 48 o 72 horas; una fila vacía es «pendiente», no cero.
  5. Pasar los ganadores al feed: el que supera la vara (1.000 de alcance en el ejemplo) vuelve al feed sin cambios y entra a un ciclo de reposteo espaciado, respetando tu tope diario.
  6. Aplicar la regla del ganador: cuando uno rompe la vara, se frena la tanda, se estudia qué tiene en común con los otros ganadores y se sube la vara (de 1.000 a 10.000).
  7. Rellenar el banco: cada reel nuevo del feed que supera tu mediana entra al banco, así el carril se alimenta solo.

#Cómo aplicarla a tu negocio

Imagina una tienda de café de especialidad en Medellín que publica tres reels por semana y tiene un puñado que funcionó muy bien (el de cómo preparar café en prensa francesa, el del tueste en vivo), pero el resto se queda en pocas vistas y no tiene tiempo para producir más.

Así arma su carril de distribución con la skill:

  1. Exporta el alcance de 90 días. Desde su herramienta de analítica saca la lista de reels con su alcance y se la pasa a Claude.
  2. Arma el banco de 30. La skill ordena, descarta los repetidos y anota el alcance original de cada uno para comparar después.
  3. Prepara cada pieza. Vuelve a exportar cada video sin tocarlo y Claude escribe un caption nuevo para cada uno, con la misma palabra de cierre que usa en el feed.
  4. Programa 5 por día en Metricool. Repartidas a lo largo del día, de mayor a menor alcance.
  5. Lee y decide a los tres días. Los que pasan de 1.000 de alcance vuelven al feed en el primer día con lugar libre. Cuando uno explota, frena las tandas y mira qué comparte con los otros ganadores antes de seguir.

Sin grabar un solo video nuevo, la tienda consigue un flujo de alcance frente a gente que no la conoce. Un aviso antes de volver a subir un ganador al feed tal cual: en nuestras cuentas, repetirlo frente a los mismos seguidores nos hizo perder alcance, y lo contamos en la regla 3 para medir contenido. Por eso conviene que el reposteo en el feed sea espaciado, como pide la skill. Para otras herramientas, mira las skills de Claude para Instagram.

#Cómo instalarla

Es una skill agent-agnostic: el mismo SKILL.md corre en cualquier agente.

  • Con una línea, para cualquier agente: npx skills add ericosiu/ai-marketing-skills --skill trial-reel-distribution -g la deja instalada en todos tus proyectos.
  • A mano: copia la carpeta trial-reel-distribution/ del repo a ~/.claude/skills/ para Claude Code, o a ~/.agents/skills/ para Cursor, Codex y Gemini CLI (las tres leen esa carpeta).
  • App de Claude: comprime la carpeta en un ZIP y súbela desde Customize, Skills, «Upload a skill».

No pide claves: necesita tu exportación de alcance por publicación y una herramienta para programar reels de prueba. El SKILL.md completo (con la licencia MIT y la fuente) está más abajo.

El SKILL.md completo

Skill original de Single Grain (Eric Siu) (MIT). Fuente: github.com/ericosiu/ai-marketing-skills/tree/main/trial-reel-distribution.

El archivo original está en inglés y va tal cual, sin traducir. Lo de arriba de esta ficha es el resumen en español: qué hace, cómo aplicarla y cómo instalarla.

Leer el archivo (97 líneas)

Trial Reel Distribution

Trial reels are shown to non-followers. That makes them a distribution channel for content that already works, not a lab for content that has never been tested. This skill runs trials that way: proven feed winners, reposted unchanged, in front of a new audience.

Principle

  • A trial slot is reach in front of people who do not follow you. Spending it on an unproven video wastes the slot: most unproven content fails in the feed too, so the trial only re-confirms the failure to a colder audience.
  • Variant testing answers the wrong question. Five rounds of edit variants on unproven masters typically move the best result from low to slightly less low, because the bottleneck is the idea, not the edit. Distribution skips the question: the feed already ran the test, and reach is the result.
  • The video is the proven part. Change anything about it and you are testing again. The only new elements are the caption and the export.
  • Winners compound. A trial that travels gets graduated to the feed, and every new feed post that proves itself becomes fresh stock for the pool.

When to use

  • You post reels to a feed regularly and have at least a few months of posts with reach data.
  • Your trial reels underperform and you have been using them to test edits, hooks, or unproven clips.
  • You want a repeatable non-follower reach lane that does not require producing new content.

Inputs you need

  1. Your feed posts from the last 90 days, ranked by reach (your analytics tool's per-post reach export is enough).
  2. The original video files for your top posts, or a way to download the finished posts back.
  3. A scheduler/analytics tool that can schedule trial reels and report per-post reach (for example Metricool, or the native app plus manual tracking). Trials do not count against your feed posting cadence.
  4. A graduation bar. 1,000 reach per trial is the worked setting for a mid-size account; scale it so only genuinely proven posts clear it on your account.

Workflow

Step 1: Build the pool
  • Rank your feed posts by reach over the last 90 days. Take the top 30 or so. These are proven: each one already earned its reach once.
  • Exclude anything you have already run as a distribution trial.
  • Record each pool item's original reach. You will compare the trial result against it later.
Step 2: Prepare each post
  • Keep the video content exactly as posted. No speed changes, no music swaps, no new hook, no re-edit.
  • Re-export the file so the upload is a fresh file, not a byte-identical duplicate of the original post. Same content, new export. Byte-identical re-uploads trip duplicate detection on both the platform and most schedulers.
  • Write a fresh caption: new wording, same topic, and the same call-to-action rules you use on the feed. Do not copy the original caption verbatim.
Step 3: Schedule in batches
  • Post about 5 trials per day, spread across the day rather than in one burst.
  • Work down the pool in reach order, best first. The strongest proven posts deserve the first slots.
Step 4: Read the results
  • Read reach per post at 48 to 72 hours. Reach is the metric; likes and comments are secondary.
  • A post whose analytics row is still empty at read time is pending, not zero. Recheck it on the next pass before judging it.
  • Log each trial's reach next to its original feed reach.
Step 5: Graduate the winners
  • Any trial at or over your graduation bar (1,000 reach in the worked example) is a winner.
  • Post the winner to your main feed, unchanged, and put it into your repost lifecycle: spaced repeat touches over the following months instead of a one-off.
  • Respect your feed cadence cap (for example 2 feed posts per day). If graduation day is full, the winner posts on the first open day. Queue discipline beats same-day posting.
Step 6: Apply the winner rule
  • When a trial breaks the bar, pause the batch flow and re-optimize around that winner first: it is the template for the next level (for example, study what its topic and format share with your other over-bar posts).
  • Then raise the bar and resume. The worked ladder is 1,000 reach first, then 10,000.
  • Do not let batch timing decide this silently. The pause is a decision point, not a scheduling accident.
Step 7: Refill the pool
  • Every new feed post that proves itself (reach clearly above your recent median) joins the pool.
  • The lane feeds itself: feed winners become trials, trial winners return to the feed with a lifecycle, and new originals keep entering at the top.

Worked example (anonymized)

One account switched its trial program from variant testing to distribution:

  • Before: 5 rounds of variant testing on unproven videos. Best single variant across all rounds: 366 reach.
  • After: the top feed posts of the last 90 days reposted as trials, video unchanged, fresh caption and export.
  • First 8 distribution posts: 5 of 8 cleared 1,000 reach. Top post: 6,107 reach. Total: about 16,900 reach.
  • Per post, distribution delivered roughly 11x the reach of the variant-testing approach.
  • The first bar-breaker triggered the winner rule: batches paused, the winner graduated to the feed, and the next bar was set at 10,000 reach.
  • Distribution then restarted as a standing lane with the rest of the pool, not as another one-off test.

Pitfalls

  • Testing unproven content in trials. The most common failure mode. If it has not proven itself in the feed, the trial slot will not save it.
  • Editing the video on the repost. The video is the proven asset. A new hook, a new cut, or a music swap turns distribution back into testing.
  • Uploading the byte-identical file. Re-export first. Duplicate guards reject identical files, and identical posts underperform.
  • Judging on a pending analytics row. Empty means not synced yet. Wait one more cycle.
  • Flooding the feed with graduates. Winners queue behind your daily feed cap. A winner posted on the first open day loses nothing.
  • Optimizing the edit when the idea is the bottleneck. If your variants cluster at low reach, stop variant testing and check whether the underlying posts ever proved themselves anywhere.
Copiar el archivo tal cual
---
name: trial-reel-distribution
description: Use Instagram trial reels as a distribution channel instead of a testing lab. Rank your feed posts by reach over the last 90 days, take the proven winners, and repost them as trial reels with the video unchanged - only a fresh caption and a fresh export. Winners that clear the reach bar graduate to the main feed. Use when asked to: "run trial reels", "distribute my best posts as trials", "repost winners as trial reels", "trial reel strategy", "why are my trial reels not working".
---

<!--
Skill original: trial-reel-distribution, de Single Grain (Eric Siu) (https://github.com/ericosiu/ai-marketing-skills/tree/main/trial-reel-distribution)
Licencia: MIT. Reproducida con atribución según los términos de la licencia.
Copia publicada en claura-ai.com/skills.
-->

# Trial Reel Distribution

Trial reels are shown to non-followers. That makes them a distribution channel for content that already works, not a lab for content that has never been tested. This skill runs trials that way: proven feed winners, reposted unchanged, in front of a new audience.

## Principle

- A trial slot is reach in front of people who do not follow you. Spending it on an unproven video wastes the slot: most unproven content fails in the feed too, so the trial only re-confirms the failure to a colder audience.
- Variant testing answers the wrong question. Five rounds of edit variants on unproven masters typically move the best result from low to slightly less low, because the bottleneck is the idea, not the edit. Distribution skips the question: the feed already ran the test, and reach is the result.
- The video is the proven part. Change anything about it and you are testing again. The only new elements are the caption and the export.
- Winners compound. A trial that travels gets graduated to the feed, and every new feed post that proves itself becomes fresh stock for the pool.

## When to use

- You post reels to a feed regularly and have at least a few months of posts with reach data.
- Your trial reels underperform and you have been using them to test edits, hooks, or unproven clips.
- You want a repeatable non-follower reach lane that does not require producing new content.

## Inputs you need

1. Your feed posts from the last 90 days, ranked by reach (your analytics tool's per-post reach export is enough).
2. The original video files for your top posts, or a way to download the finished posts back.
3. A scheduler/analytics tool that can schedule trial reels and report per-post reach (for example Metricool, or the native app plus manual tracking). Trials do not count against your feed posting cadence.
4. A graduation bar. 1,000 reach per trial is the worked setting for a mid-size account; scale it so only genuinely proven posts clear it on your account.

## Workflow

### Step 1: Build the pool

- Rank your feed posts by reach over the last 90 days. Take the top 30 or so. These are proven: each one already earned its reach once.
- Exclude anything you have already run as a distribution trial.
- Record each pool item's original reach. You will compare the trial result against it later.

### Step 2: Prepare each post

- Keep the video content exactly as posted. No speed changes, no music swaps, no new hook, no re-edit.
- Re-export the file so the upload is a fresh file, not a byte-identical duplicate of the original post. Same content, new export. Byte-identical re-uploads trip duplicate detection on both the platform and most schedulers.
- Write a fresh caption: new wording, same topic, and the same call-to-action rules you use on the feed. Do not copy the original caption verbatim.

### Step 3: Schedule in batches

- Post about 5 trials per day, spread across the day rather than in one burst.
- Work down the pool in reach order, best first. The strongest proven posts deserve the first slots.

### Step 4: Read the results

- Read reach per post at 48 to 72 hours. Reach is the metric; likes and comments are secondary.
- A post whose analytics row is still empty at read time is pending, not zero. Recheck it on the next pass before judging it.
- Log each trial's reach next to its original feed reach.

### Step 5: Graduate the winners

- Any trial at or over your graduation bar (1,000 reach in the worked example) is a winner.
- Post the winner to your main feed, unchanged, and put it into your repost lifecycle: spaced repeat touches over the following months instead of a one-off.
- Respect your feed cadence cap (for example 2 feed posts per day). If graduation day is full, the winner posts on the first open day. Queue discipline beats same-day posting.

### Step 6: Apply the winner rule

- When a trial breaks the bar, pause the batch flow and re-optimize around that winner first: it is the template for the next level (for example, study what its topic and format share with your other over-bar posts).
- Then raise the bar and resume. The worked ladder is 1,000 reach first, then 10,000.
- Do not let batch timing decide this silently. The pause is a decision point, not a scheduling accident.

### Step 7: Refill the pool

- Every new feed post that proves itself (reach clearly above your recent median) joins the pool.
- The lane feeds itself: feed winners become trials, trial winners return to the feed with a lifecycle, and new originals keep entering at the top.

## Worked example (anonymized)

One account switched its trial program from variant testing to distribution:

- Before: 5 rounds of variant testing on unproven videos. Best single variant across all rounds: 366 reach.
- After: the top feed posts of the last 90 days reposted as trials, video unchanged, fresh caption and export.
- First 8 distribution posts: 5 of 8 cleared 1,000 reach. Top post: 6,107 reach. Total: about 16,900 reach.
- Per post, distribution delivered roughly 11x the reach of the variant-testing approach.
- The first bar-breaker triggered the winner rule: batches paused, the winner graduated to the feed, and the next bar was set at 10,000 reach.
- Distribution then restarted as a standing lane with the rest of the pool, not as another one-off test.

## Pitfalls

- Testing unproven content in trials. The most common failure mode. If it has not proven itself in the feed, the trial slot will not save it.
- Editing the video on the repost. The video is the proven asset. A new hook, a new cut, or a music swap turns distribution back into testing.
- Uploading the byte-identical file. Re-export first. Duplicate guards reject identical files, and identical posts underperform.
- Judging on a pending analytics row. Empty means not synced yet. Wait one more cycle.
- Flooding the feed with graduates. Winners queue behind your daily feed cap. A winner posted on the first open day loses nothing.
- Optimizing the edit when the idea is the bottleneck. If your variants cluster at low reach, stop variant testing and check whether the underlying posts ever proved themselves anywhere.

Preguntas frecuentes sobre la skill trial-reel-distribution

¿Qué es la skill trial-reel-distribution y para qué sirve?

trial-reel-distribution es una skill de Single Grain, la agencia de Eric Siu (github.com/ericosiu/ai-marketing-skills, licencia MIT), que cambia cómo usas los reels de prueba de Instagram. Como esos reels se muestran a gente que no te sigue, propone usarlos para distribuir lo que ya funcionó en tu feed y no para probar videos nuevos. Arma el banco de ganadores, el ritmo de publicación, la lectura de resultados y la regla para pasar al feed los que superan tu vara.

¿Cómo la aplico en mi cuenta?

Ordenas tus reels de los últimos 90 días por alcance y te quedas con unos 30. Cada uno se vuelve a subir como reel de prueba con el video idéntico, pero exportado de nuevo y con un caption distinto. Subes unas 5 pruebas por día, lees el alcance a las 48 o 72 horas y el que supera tu vara vuelve al feed. Para hacerlo con Claude paso a paso, mira cómo publicar trial reels con Claude.

¿Por qué no conviene usar los reels de prueba para testear ediciones?

Según la skill, porque casi nunca el problema es la edición sino la idea. En el ejemplo que documenta, cinco rondas de variantes sobre videos sin probar dieron como máximo 366 de alcance, y los primeros 8 reels redistribuidos (ganadores del feed, sin editar) sumaron unos 16.900, con 5 de 8 por encima de 1.000. Si cambias el video, vuelves a estar probando: lo único nuevo debe ser el caption y la exportación.

¿Qué necesito para usarla y sirve en español?

Necesitas una cuenta de Instagram con varios meses de reels y sus datos de alcance (la exportación de tu herramienta de analítica alcanza), los archivos originales y una forma de programar reels de prueba, como Metricool o la app de Instagram con registro a mano. No pide claves. Es un método, no depende del idioma: le pides los captions nuevos en español y aplica las mismas reglas.

Más skills de Redes sociales

Redes sociales

ig-carousel-planner: planificar carruseles de Instagram

Planifica un carrusel de Instagram lámina por lámina: elige la fórmula según el objetivo, escribe el texto de cada imagen y el caption, y deja la última lámina lista para que la guarden.

SkillPrincipianteSKILL.md libre
Redes sociales

ig-hook-extractor: descifrar el hook de un viral

Desarma el gancho de un reel o carrusel viral de Instagram: qué fórmula usa, por qué funcionó, qué buscaba y una plantilla en blanco para tu tema.

SkillPrincipianteSKILL.md libre

Para seguir con este tema

Elegí qué cookies usamos