DEV Community: Support ExpatHub

I Built a YouTube Analytics Tool After Spending 2 Hours Not Understanding Why My Video Flopped

Support ExpatHub — Tue, 23 Jun 2026 08:43:42 +0000

That's how long I spent staring at YouTube Studio after a video
got half the views I expected. I clicked through every tab.
Impressions. Reach. Retention. Traffic sources.

I saw the numbers. I had no idea what they meant.

Not because I'm bad at analytics. Because YouTube Studio gives
you data without interpretation. It tells you what happened.
Never why.

The benchmark problem nobody talks about

After digging into this for months, I found one mistake that almost
every creator makes.

They compare their CTR to YouTube's platform average.

You've probably seen the advice. "A good CTR is 4-6%."
Some people say 2-3% is fine. Others say aim for 8%.

All of it is noise.

A documentary channel and a gaming channel have completely different
audience behaviors, different thumbnail styles, different viewer
expectations. Comparing them to the same benchmark tells you
absolutely nothing useful.

The only number that actually matters is your own channel's
historical average. How did this video perform compared to
your last 20 videos? That's the question.

What I started tracking instead

Once I stopped looking at platform benchmarks and started comparing
each video to my own channel baseline, patterns became obvious.

Some things worth tracking:

CTR vs your own average. Not YouTube's suggested range.
Your personal average from the last 20-30 uploads. A video
at 3.2% CTR is either great or terrible depending entirely
on your channel's baseline.

Retention at the 30-second mark. If viewers are leaving
before 30 seconds, YouTube stops recommending the video.
This single metric explains more failed videos than anything else.

Subscriber conversion per video. Some videos get decent
views but convert zero new subscribers. That tells you the
content is attracting the wrong audience, not that the video itself failed.

Browse Features traffic percentage. When this is low,
the algorithm isn't pushing your video to non-subscribers.
High impressions but low browse traffic means YouTube
tried but the thumbnail didn't click.

Why I built CreatorPilot

Tracking all of this manually for every video is exhausting.
After doing it in spreadsheets for a while, I built a tool
to automate it.

CreatorPilot connects directly to YouTube Analytics API,
calculates your channel's own performance baseline, and uses
AI to explain why each video over or underperformed.
Not generic tips. Actual diagnosis based on your specific numbers.

It also matches your channel category to what's currently
trending, so the next video recommendations are based on
real data rather than guesses.

Still early, free plan available at creatorpilothq.com.

If you're a creator spending hours in Studio with more
questions than answers, it was built for exactly that.

I built an AI tool that explains why YouTube videos underperform

Support ExpatHub — Sun, 21 Jun 2026 08:48:24 +0000

Spent the past few months building CreatorPilot, a tool that connects to
YouTube Analytics API and uses Claude to diagnose why specific videos
underperformed.

The core problem I was solving for myself: YouTube Studio gives you tons
of data but zero interpretation. You can see your CTR dropped, but not
why, or what to actually do about it.

Tech stack:
Next.js for the frontend
YouTube Analytics API + YouTube Data API for the data layer
Claude API for the analysis and recommendations
Stripe for billing

The interesting technical challenge was getting the AI analysis to be
useful instead of generic. Early versions just said things like "consider
improving your thumbnail" for every low-CTR video, which is technically
true but useless. Had to build in actual comparison logic against the
channel's own historical baseline before passing data to the model, so
the output references real numbers instead of generic advice.

Currently supports 5 languages (EN/KO/JP/DE/ZH) since YouTube creators are
obviously a global audience and most analytics tools are English-only.

Free plan available if anyone wants to poke around. Would love feedback
from other devs, especially on how you've handled the "AI sounds confident
but might be wrong" problem in your own projects.

creatorpilothq.com

I Built an AI Tool That Tells YouTube Creators Why Their Videos Fail

Support ExpatHub — Fri, 19 Jun 2026 09:03:46 +0000

The Problem

Every YouTube creator has had this experience: you spend days on a video,
upload it, and it gets half the views you expected. You open YouTube Studio,
stare at the numbers, and have no idea why.

View count is down. But why?

Was it the thumbnail? The title? Did the algorithm stop pushing it?
Was the topic just bad timing?

YouTube Analytics gives you data. It doesn't give you answers.

What I Built

CreatorPilot connects directly to your YouTube Analytics API and uses
Claude AI to actually explain what happened.

Here's what it does differently:

1. Grades each video against YOUR baseline — not other creators
A video with 5,000 views might be a flop for one channel and a hit for another.
CreatorPilot compares each video to your own channel average.

2. Explains WHY — not just what
Instead of showing you a CTR number, it tells you:
"This video's CTR dropped 40% below your average — likely a thumbnail issue,
not a topic issue."

3. Matches your channel to what's trending RIGHT NOW
It pulls live YouTube trending data and finds topics that fit
your specific channel category and style.

4. Predicts monetization timeline
Based on your actual subscriber growth rate and watch time, it calculates
how many days until you hit YouTube monetization requirements.

Tech Stack

Next.js + TypeScript
Claude Sonnet (AI analysis)
YouTube Analytics API v2
YouTube Data API v3
Google Trends API
Stripe (payments)
PostgreSQL + Prisma

Try It Free

→ creatorpilothq.com

Free plan available. No credit card required.

Would love feedback from other builders — especially on the AI analysis side.

How I Built an AI YouTube Analytics SaaS with Next.js and Claude API (and kept costs under $2/user/month)

Support ExpatHub — Thu, 18 Jun 2026 09:34:54 +0000

I recently launched CreatorPilot, an AI-powered YouTube analytics platform. Here's how I built it and the technical decisions I made.

The Problem

Most YouTube creators manually copy their analytics into ChatGPT to figure out why their videos perform the way they do. It's tedious and the advice is generic because ChatGPT doesn't actually know your channel data.

The Solution

Connect directly to YouTube Analytics API and feed real channel data into Claude AI to generate specific, actionable insights.

Tech Stack

Frontend/Backend: Next.js 14 (App Router)
Database: PostgreSQL + Prisma ORM (Neon)
AI: Anthropic Claude API
Auth: NextAuth.js with Google OAuth
Payments: Stripe
Deployment: Vercel

The Cost Optimization Challenge

Claude Sonnet costs $3/1M input tokens. With 5 AI features per user, naive implementation would cost $5-7/user/month — killing margins.

Solution: Aggressive DB caching

Feature	Cache Duration	Model
Video Analysis	24 hours	Haiku
Daily Trend	24 hours	Haiku
Action Plan	24 hours	Haiku
20 Video Recs	170 hours	Haiku
Channel Pattern	170 hours	Sonnet

Result: ~$1.35/user/month worst case, ~$0.41 average. Gross margin ~90%.

Key Technical Challenges

1. YouTube API Quota Management
YouTube Data API has a 10,000 quota/day limit. Solved by caching video data in PostgreSQL and only calling the API when cache expires.

2. Multilingual AI Responses
Supporting 5 languages (EN, KO, JA, DE, ZH) with consistent quality required careful prompt engineering — especially for Korean historical content analysis.

3. Rate Limiting Without Redis
Built a PostgreSQL-based rate limiting fallback when Redis is unavailable.

Results

Launched today. Would love feedback from fellow developers.

https://creatorpilothq.com