[Journey] Building AI content sites for SEO ($1600 MRR) — A Software Engineer's perspective

[kingkhavn]

My journey building AI content sites for SEO — a software engineer's perspective​

---

Background: I'm a software engineer. CS degree, been working in AI/ML for a while. A few years ago I got into SEO almost by accident — started applying the same engineering mindset to content. Build, measure, iterate.

That led to building a lot of travel sites. Over 100 at this point.

Most are dead. One works really well — peaked over 1k clicks/day recently. A few are somewhere in the middle. Revenue is around 1600$ per month





I'll share GSC screenshots as this thread goes on.

---

Why travel

Honestly, it was a good fit for what I was trying to do technically. Travel content is highly structured — attractions, prices, opening hours, itineraries. Easy to model, easy to generate at scale, easy to validate.

It's also brutally competitive. Which made it a good stress test.

---

What I actually built

I didn't just publish AI articles and wait. That part I figured out pretty early doesn't scale the way people think it does.

What I ended up building is closer to a content ops system:

— daily GSC monitoring per page, automated
— a scoring function that decides which pages to touch and how
— an autopilot that commits changes (content refresh, meta rewrite, internal links) based on that score
— a recheck loop at 5–7 days to confirm or revert

The content is AI. But the decision layer on top of it is engineered, not vibes-based.

That distinction matters a lot. I'll get into the specifics as the thread goes on.

---

What this thread is

Just a log. I'll post what I'm testing, what the numbers say, what broke, what worked.

No course. No upsell. I just think this space moves fast enough that it's worth documenting in public.

If you're running AI content sites and thinking about how to systematize it — follow along. And if you're already doing something similar, I'd genuinely like to compare notes.

---

More coming soon — starting with the GSC data.

 

[kingkhavn]

---

The data — 6 months, 3 sites

Let me show you what this actually looks like before I explain anything.

137K clicks, 22.3M impressions over 6 months. Avg position 12.2.

The curve tells the real story though. Grew steadily from October, peaked around early January at nearly 1,700 clicks in a single day — then dropped hard. That drop wasn't a penalty. It was seasonality. European travel traffic tanks in January. I know that now.

It's been recovering since February and is currently stabilizing around 500–600/day. The impressions line (lighter) is tracking back up ahead of the clicks line — that's usually a good sign.

This is the site the system is built around. 22,000+ pages, automated monitoring, daily commits.

Site 2 — the one that got away

12.9K clicks, 1.71M impressions. Avg position 15.2.

This one peaked around 160 clicks/day back in late October, then fell steadily for four months straight. Still hasn't recovered. Currently sitting at maybe 20–30/day.

I didn't have any monitoring on this when it started dropping. By the time I noticed, the decline was already baked in. That's the version of "publish and move on" that costs you — you don't even know when something starts dying.

I've started running the system on this one now. Slight uptick in impressions the last few weeks. Too early to call it a recovery but it's not getting worse.

Site 3 — early stage, just waking up

879 clicks, 120K impressions. Avg position 18.2.

This one is interesting. Basically flatlined from October through February — single digit clicks most days. Then something clicked in March and it spiked to 25–30 clicks/day. Still small but the shape of that curve is exactly what you want to see on a new site.

18 posts. No real automation yet. First signal from a cross-link injection I ran from the main site in late March. Watching it closely.

---

These three charts basically represent the three phases of how I've approached this:

— Site 1: actively managed with a system
— Site 2: built and abandoned, now trying to recover
— Site 3: early stage, just starting to get attention

More detail on what I actually did differently on each — next update.

 

[clau777]

Following. Also with 3 sites, not getting the traffic you have but growing

 

[bottingisfine]

I am literally on the same boat.

How many pages do you build per domain?

Do you do any link building (if so, which kind?) or just content?

 

[Patrick Seo1]

Total for ads, how much bro

 

[GroByTech]

Why didn't you use AdSense? Is there any specific reason, for example Adsense rejected because of AI(thin content)?

 

[kingkhavn]

I am literally on the same boat.

How many pages do you build per domain?

Do you do any link building (if so, which kind?) or just content?

i tested lot of numbers ( 20k, 10k, 5k, 250, 50). link building tested too ( most with rankerx)

 

[kingkhavn]

Total for ads, how much bro

No ads.

Why didn't you use AdSense? Is there any specific reason, for example Adsense rejected because of AI(thin content)?

yes, i got rejected. but with my calculation, affiliate performance better ( because it's tailored specified for content, not random)

 

[GroByTech]

No ads.


yes, i got rejected. but with my calculation, affiliate performance better ( because it's tailored specified for content, not random)

Your journey is an inspiration to many who think only about the AdSense revenue. In one way, if someone gets traffic from social platforms like Reddit, it can make a good amount of income

 

[RIA 377]

Without incurring advertising costs, this is a journey worth following and learning from. Have you built a system of articles promoting tourist destinations without spending money?

 

[TusherSEO]

My journey building AI content sites for SEO — a software engineer's perspective​

---

Background: I'm a software engineer. CS degree, been working in AI/ML for a while. A few years ago I got into SEO almost by accident — started applying the same engineering mindset to content. Build, measure, iterate.

That led to building a lot of travel sites. Over 100 at this point.

Most are dead. One works really well — peaked over 1k clicks/day recently. A few are somewhere in the middle. Revenue is around 1600$ per month

View attachment 515500

View attachment 515501

I'll share GSC screenshots as this thread goes on.

---

Why travel

Honestly, it was a good fit for what I was trying to do technically. Travel content is highly structured — attractions, prices, opening hours, itineraries. Easy to model, easy to generate at scale, easy to validate.

It's also brutally competitive. Which made it a good stress test.

---

What I actually built

I didn't just publish AI articles and wait. That part I figured out pretty early doesn't scale the way people think it does.

What I ended up building is closer to a content ops system:

— daily GSC monitoring per page, automated
— a scoring function that decides which pages to touch and how
— an autopilot that commits changes (content refresh, meta rewrite, internal links) based on that score
— a recheck loop at 5–7 days to confirm or revert

The content is AI. But the decision layer on top of it is engineered, not vibes-based.

That distinction matters a lot. I'll get into the specifics as the thread goes on.

---

What this thread is

Just a log. I'll post what I'm testing, what the numbers say, what broke, what worked.

No course. No upsell. I just think this space moves fast enough that it's worth documenting in public.

If you're running AI content sites and thinking about how to systematize it — follow along. And if you're already doing something similar, I'd genuinely like to compare notes.

---

More coming soon — starting with the GSC data.

This is a very impressive journey, mate! Coming from a software engineering background definitely gives you a unique edge in scaling AI content. 100+ sites is a massive number!

I'm curious about the sites that are 'somewhere in the middle'—are you planning to optimize their technical SEO or focus more on building fresh backlinks to boost them? Following your progress closely!

 

[BuildMoreLinks]

how is the revenue $1600 with no ads? Also what AI model are you using for the content. ? How old are these sites?

 

[kingkhavn]

Without incurring advertising costs, this is a journey worth following and learning from. Have you built a system of articles promoting tourist destinations without spending money?

Ofcourse there are costs for LLM, hostings, domain, images, backlinks but i always try to minimize it. And yes i have full pipeline

 

[kingkhavn]

This is a very impressive journey, mate! Coming from a software engineering background definitely gives you a unique edge in scaling AI content. 100+ sites is a massive number!

I'm curious about the sites that are 'somewhere in the middle'—are you planning to optimize their technical SEO or focus more on building fresh backlinks to boost them? Following your progress closely!

I’m focus on technical seo to rescue them.

how is the revenue $1600 with no ads? Also what AI model are you using for the content. ? How old are these sites?

I’m using gemini mainly, but now trying to switch to claude. Some sites more than 1 year old, some news. I try continously

 

[kingkhavn]

By the way, dead sites somehow are just dead for Google; other search engines still send traffic.

 

[spectrejoe]

Since you can do things at scale it would be interesting to test and try to pinpoint what's causing google to blacklist / kill some of your sites.

 

[kingkhavn]

Since you can do things at scale it would be interesting to test and try to pinpoint what's causing google to blacklist / kill some of your sites.

I’m trying to increase my success rate. If I already knew what would work, I’d already be rich.

 

[CTRPhantom]

Nice work. That content-ops setup is exactly how you scale. How do you keep 22k+ pages fresh without burning quality? Any tips on avoiding drops while staying fast?

 

[Starblazer]

Have you tried to identify the root cause of traffic growth or decline like niche, thin content, site design, backlinks, etc.? If you figure out what works and do some testing, you can easily scale it up.

You can use multi-quote feature to respond to multiple comments at once. Just click "+Quote" button to quote a comment to your response.

 

[Elevate Studio]

Solid share

The decision layer point is key.