YouTube Thumbnail A/B Tester

What it does

Analyzes an existing thumbnail's weaknesses, proposes 2-3 variants each tied to a specific hypothesis (contrast, face expression, text size), and prepares them for YouTube's native Test & Compare tool. Reports results only from YouTube's real data, never estimated CTR.

When to use it

• A video's click-through rate is underperforming and you want a structured test instead of guessing
• You're rolling out a new thumbnail style and want proof before applying it channel-wide
• You manage several channels and need a repeatable experimentation process

Steps

1. Pull the current thumbnail and its real CTR from YouTube Studio (no estimates)
2. Write one clear hypothesis per variant (e.g. 'larger face increases CTR')
3. Design each variant so only the hypothesis element changes
4. Generate the variant images at 1920x1080 and QA them for cropping and legibility
5. Load variants into YouTube's Test & Compare tool and set the test duration
6. Report real results once the test concludes and recommend the winner for approval before making it permanent

The skill

---
name: youtube-thumbnail-ab-tester
description: Design hypothesis-driven thumbnail variants and run them through YouTube's A/B test tool to find the highest-CTR option.
---

# YouTube Thumbnail A/B Tester

## 1. Overview
Analyzes an existing thumbnail's weaknesses, proposes 2-3 variants each tied to a specific hypothesis (contrast, face expression, text size), and prepares them for YouTube's native Test & Compare tool. Reports results only from YouTube's real data, never estimated CTR.

## 2. When to Use
- A video's click-through rate is underperforming and you want a structured test instead of guessing
- You're rolling out a new thumbnail style and want proof before applying it channel-wide
- You manage several channels and need a repeatable experimentation process

## 3. Steps
1. Pull the current thumbnail and its real CTR from YouTube Studio (no estimates)
2. Write one clear hypothesis per variant (e.g. 'larger face increases CTR')
3. Design each variant so only the hypothesis element changes
4. Generate the variant images at 1920x1080 and QA them for cropping and legibility
5. Load variants into YouTube's Test & Compare tool and set the test duration
6. Report real results once the test concludes and recommend the winner for approval before making it permanent

## 4. GoHighLevel Angle
Apply this inside GoHighLevel workflows (funnels, automations, or client reporting) wherever the same task recurs for an agency or client account.

## 5. Safety Rules
- Always produce a draft first; never send, publish, or spend money without explicit human approval.
- Never invent data, metrics, or results — use only what is confirmed or clearly label estimates.
- Never paste API keys, tokens, or passwords into chat; reference secrets by name from a secure store.

For Claude Code, save it as ~/.claude/skills/youtube-thumbnail-ab-tester/SKILL.md. In Cursor, ChatGPT or another assistant, paste it as project instructions.