What running a 1.3M-user site solo taught me about engineering

Jun 2, 2026
BlogEngineeringPerformanceSolo

No team. No on-call. Just me and a French Bulldog blog that pays the bills.

TL;DR: I built a French Bulldog blog to 1.3M annual readers and six figures a year, by myself. Running it solo taught me more about engineering than any job has: kill the polish, find the two levers that actually move money, treat speed and uptime as revenue, and build things solid enough that you stop fearing the update. The long version is below.

Contents

I started FrenchieWiki in 2019.

For two years it crawled. Slow to grow, slow to rank, slow to make a dollar. I almost quit more than once.

Then the curve bent. Flat line to hockey stick, basically overnight. Today the site reaches over 1.3 million people a year and earns six figures from display ads.

Still just me.

When it breaks, I’m the one who fixes it. I’m also the one whose income just dropped. That second part teaches you things a salary never will.

Polish is a trap

I used to tweak forever. A margin here. A hover state there. Some component buried three clicks deep that maybe forty people ever saw.

Then I’d open the analytics. The thing I’d burned an afternoon on had been seen by 0.0001% of users.

That number cured me.

When your time is free, polishing feels like work. When you’re solo and every hour costs you real money, one question shows up before you touch anything: does this move the needle, or does it just feel good?

Usually it just feels good.

Two things mattered. The rest was noise.

Content and speed. That was the whole list.

Good pages and fast pages moved revenue. The gorgeous UI I kept fussing over moved nothing. Flat. You only learn that by being wrong with your own money, over and over, until you stop arguing with the numbers.

Speed is a feature you can put a price on

Since speed paid, I built for speed.

The stack was boring on purpose. Static Gatsby build, hosted on Netlify, Cloudflare sitting in front.

Cloudflare’s CDN and tiered cache handed most visitors a copy from a server near them. The origin barely got touched. Then the basics, done without mercy: tight cache rules, every CSS and JS file trimmed, images crushed and served as WebP, every free win switched on.

None of it was clever. But when a slow page costs you rankings, and rankings cost you money, you stop calling performance “an optimization.” You start calling it the job.

Uptime is just revenue wearing a different hat

I’ll be honest. I’ve never fought a real fire. The system is simple, and simple things don’t break much.

I learned the lesson sideways instead. Ads blinking offline. A few minutes of downtime here and there. Rankings dipped every single time. Small, temporary, and completely real. The site would come back and the cost would hang around for days.

You feel reliability differently when the dashboard is your bank balance. You quit over-engineering for failures that never come, and you start respecting the few that do.

I never feared Google updates

People in my niche panic every time Google ships an algorithm update. Forums light up. Everyone braces for impact.

I didn’t.

I’d built the site out of the exact things those updates reward. Real content. Fast pages. No tricks. There was nothing there for an update to punish.

Build it right and you stop fearing the thing everyone else fears. The audit. The update. The traffic spike. The dread other people carry usually means they cut a corner and they’re praying nobody checks.

What this did to me as an engineer

My day job is software for regulated healthcare systems. Different world. Higher stakes, more process. The reflexes from the blog came with me anyway.

I ask whether the work matters before I do it. I treat speed and uptime as features with a price tag, not chores for later. I can scan a roadmap and usually pick the one item that compounds out of the nine that are busywork. I build things solid enough that I’m not nervous about what hits them next month.

No job taught me that.

Being the only one who could fix it did.

×