[Journey] Ranking AI Sites For Passive Income

So I've had quite a few journeys here before, majority successful, some I simply stopped half way or continued without updating but for the most part I've been working in IM for the past 15 years so yea pretty successful I would say.

I actually haven't done much SEO for the past couple years as I settled into a niche more about software development where I barely gotta do fuck all besides maybe once a month. However, I been kinda bored and with a lot of free time on my hands so decided why not go back to my roots?

What I've done so far:
- Bought Domain
- Decided to go with a general website with multiple niches instead of 1 website per niche (I actually got good results with the 1 website per niche but I'll give a general website a try first)
- Improved my AI Prompts to generate me banger articles (~2-3.5k Words each, extremely in-depth content)
- Okay'ish workflow to find keywords (not too happy but I can find SOME good ones, although I would prefer to have something better like Ahrefs but cba spending 120/month for that alone).
- Uploaded 9 Articles

Not all articles are meant to make me money, and I won't really go in-depth on why I make them but lets just say it's for increased topical authority. Around half the articles I post are actual "money making" articles.

I will be monetizing with Amazon Affiliate initially and if the non money making articles start popping off I'll give ads a try as well.

Plan moving forward:
- Add 2-3 articles a day
- Create automated workflow to pump out articles en mass
- Once my Article App is finished I'll increase to 10-20+ articles per day & create more websites.
- Sell websites once earnings begin to plateau (I always made the mistake of keeping them for the monthly revenue but it was rarely ever worth, better to just sell, reinvest the money)

Day 1
- Ended up working a bit more than expected for the first day.

-Added another 4 articles, all buyer intent focused / commercial, monetized with Amazon links.
These articles I created may have some keyword cannibalism going on, never actually tested so don't know if these types of really specific and laser targetted keywords will work or not but only one way to find out.

- Tweaked my prompts a little bit more to fix some edge case issues.

- Started working on my AI Bot to scale this project a little bit harder.
Figured out that even though I have ChatGPT Plus I actually don't get access to the API and it's a whole other service that I gotta buy for the API and it's paid on demand it looks like. While I'm interested I'm a cheap motherfucker so I just fucked around and figured a way to bypass the API and simply get what I need through Web version of ChatGPT. Yes it is slower, yes it will probably require more frequent updates but it's something that's easy to do and it's basically free since I already had my subscription for my every day coding needs. However the idea is that with the initial profits I will move to the actual API.

The articles themselves are looking really good and quite in-depth. They do require quite a bit of manual editing to add the Amazon links, add the amazon tables/product blocks, etc... So even if I automate the articles it will take me at least 5-10 minutes to edit each article. If I ever get to a point where this is making me more than 2K/Month I'll just hire a VA to deal with this shit. The key to all of this is automating as much as I possibly can while retaining quality.

Once I manage to scale it also means I can target non commercial keywords and go for regular ads which will not require me to touch the articles at all.

So far I made sure that my articles are extremely optimized for ranking:
- Between 2k and 3.5k words long
- More information related to the topic than competitors
- Aim to answer multiple problems that readers may have related to the main topic.
- Semantic words used to target a wide array of keywords in each article
- In Depth review of each product
- Real World Experiences / Examples (For E-E-A-T)
- External Links to Authority Websites
- Internal Linking to only highly contextual articles (won't do any if it doesn't find any highly contextual article)
- FAQ at the end to further target a few extra keywords.
- Featured Image with main keyword as alt + title
- Keyword density is always extremely low since very long article

Not sure if I'm missing anything else that's important to rank in terms of On-Page SEO, especially for E-E-A-T, will have to do some more checks but I'm already quite satisfied.

New Goals:
- Reinvest initial profits into ChatGPT API + Ahrefs (So need at least ~160$/month minimum)

I don't expect to take too long to get there, maybe a few months since brand new domain but after that it should be smooth sailing.

Good luck with your journey. Anyway, why don't you build auto generate and auto publish? or you have to maintain every generated article?

Elias19 said:

Good luck with your journey. Anyway, why don't you build auto generate and auto publish? or you have to maintain every generated article?

Click to expand...

Thats basically what I did but theres still some stuff to change and fix before its perfect, should finish it tomorrow.

But theres things that I cant automate like getting the affiliate links from amazon and stuff. The gutenberg blocks I can automate them but its gonna consume extra tokens from chatgpt which will reduce how many articles I can generate, will see what I can cook up. I dont intend to go for more than 10 articles a day since I just dont have a good enough method to find low competition keywords in mass just yet, waiting to get ahrefs for that then ill scale it even harder.

Goodluck with your journey

Good luck! I plan on starting a similar journey soon. I'm writing a program from scratch to automate a lot of the steps.

I would recommend using a serp API to get potential keywords.

End goal is to use Google search console API to feed back new keywords I start ranking for and determine if a new article should be created or if a article should be updated to encompass it.

chrischris15 said:

Good luck! I plan on starting a similar journey soon. I'm writing a program from scratch to automate a lot of the steps.

I would recommend using a serp API to get potential keywords.

End goal is to use Google search console API to feed back new keywords I start ranking for and determine if a new article should be created or if a article should be updated to encompass it.

Click to expand...

Yea I mean I know Ahrefs is really damn good for finding good keywords so that's on my plans for the future once the website reaches 100$+/month. Right now all I do is use a bunch of tools which gather keywords suggest by google autocomplete for the most part, while it helps with some keywords, it's not great for finding the really hidden ones.

GSC for finding keywords is good yea but you already need a good amount of articles to be ranking & stuff so you can see which keywords you're not in top 10 and either update articles and add them in or make new articles targetting those specific keywords.

Update #2

- Finished the first version of the Automated Workflow to generate articles, Featured Image & upload them to my wordpress site.
I'm happy with the results, it does take a good while per article (~10-20 minutes per article because they're long & im also generating really high quality images).
The whole process will have to be refactured later down the line once I move to the API but I'm extremely happy already with the current process and honestly up to 20-30 articles a day isn't bad at all if I target the right keywords, I won't even get that high anyways, maybe once I have more websites.

Took me quite a bit of finetuning the workflow and fixing edge cases and had to monitor everything it did & read every single article it pumped out to ensure the quality was perfect and to ensure that there were no issues.

- Finetuned the prompts to further optimize my articles.
Honestly, I'm extremely happy with the state of my articles, I do feel like they're getting maybe a little bit too bulky (~3k to 3.5k words per article) but I'll see how this works out for ranking.

With all the finetuning I managed to:

1. Improve the internal linking logic
2. Improve the article to make sure it always answers the users search intent right after the intro and then moves on to provide more context & information about the subject.
3. Added more personality
4. Faked real world "experiences"/situations for E-E-A-T
5. Improved automated tables for products that the AI Generated
6. Improved the way the AI seamlessly weaves internal links & external authority links with the content, it does it in a way that actually fits instead of just forcing a paragraph telling people to check the article, it's actually quite natural now.

Overall I'm extremely happy with my articles.

I also thought about getting the starter version of Ahrefs to speed up the keyword research, will probably go with that, what's 30$ anyways.

Also submitted 5 of my links to GSC to get indexed, will do 5 a day. I'm sure after a couple weeks the internal linking will do enough to ensure everything gets indexed naturally quite fast.

Site #1 (General / Multi Niche)
Posts: 24
Views: 0
Earnings: 0$

Update #3

- Bit the bullet and bought the ahrefs "starter" edition.
I was both underwhelmed and surprised at the same time. I was surprised because I actually added all my sites to it to get some feedback and ahrefs pointed out a pretty glaring issue on my main website that's making me the most money (outside of this journey). I somehow was injecting 2 <title> tags in every single page, they had the exact same title but I heard google doesn't like that. If I get even a tiny bit of a boost in rankings my earnings will skyrocekt with this site since most of my pages are in top 4-5 and the CTR boost of breaking top 3 is crazy.

So that was really good find I guess.

But I was underwhelmed with the keyword research part, it did find a bunch of keywords but it was underwhelming, the keywords it found weren't anything special and I actually already knew about most of them. The fact that I can't export all of them either due to being in the starter edition is pretty shit as well not gonna lie. Not sure if I'll keep paying for the subscription, I'll keep using it and I'll see.

- Did some reflection and decided to only post ~3 articles a day for now.
The reason being that I saw last year or two there was an update regarding article velocity, especially aimed at low authority sites & brand new websites, which makes sense really, with the rise of AI it makes sense that you don't want garbage flying around and most "new" websites shouldn't have the capacity to hire huge teams of writers to pump out dozens of articles a day.

Since I'm not in a hurry and since my articles are so bulky (2-3.5k words per article) I decided to stay with 3 articles a day for now and I will ramp up later as I gain some authority.

With that said I did decide to reduce niches and focus on only a couple for this site instead since I can only post 3 articles a day for a good while, would take too long to really get good results without building good topical authority by just shooting everywhere.

- Started Site #2
This one is in an evergreen niche, it's huge, has a TON of keywords I can target even without using Ahrefs and it's for the most part informational content (with a few commercial but not really profitable). Informational content is fine, I've never had huge earnings with them but I do know some people out there make a ton of money from ads so why not give it a try.

Bought the domain and set it up with GeneratePress Premium theme, perfectly customized the theme for my niche, got the website all setup to fulfill Google's E-E-A-T requirements and posted 4 articles (Was gonna do 3 but I wanted to test something so did a 4th).

- More small tweaks to my AI Prompts.
Made it add callout boxes to my articles, they look good, keeps the readers attention = higher reading time = better rankings.


I also noticed that my first website somehow is also already getting visitors, how or why I have absolutely no clue! But it's good news nonetheless of course.


Lets hope it keeps up like this.


Site #1
Posts: 25
Views: 24

Site #2
Posts: 4
Views: 0

Earnings: 0$

it is different from others journey and it is very difficult but I see you are managed very smoothly.

What prompt do you use for contents?

Mr Khan said:

What prompt do you use for contents?

Click to expand...

Unfortunately I can't share it, it's actually multiple prompts and not just one and it's the reason why my articles are so damn good. If I share not only would it make it easier for everyone to trace my articles & websites (because of tone, articles structure similarity, etc) but also would increase my competition. All I can say is that it pays off to be detailed in your prompts. See how big websites write their articles and try to replicate that then fine tune everything until you are satisfied.

I'm always constantly improving my prompt.

For everyone else:
I'll post a new update in a few days, for now I'm just posting daily articles, all websites are new so it takes time until google starts picking things up.

Yeah if you are generating a article with a single prompt it won't be that good.

I see a lot of people using n8n to create AI workflows.

The more you are able to give AI to guide that article the better. And with gpt-5.4 having a 1m+ context window that is great for some really well written articles.

chrischris15 said:

Yeah if you are generating a article with a single prompt it won't be that good.

I see a lot of people using n8n to create AI workflows.

The more you are able to give AI to guide that article the better. And with gpt-5.4 having a 1m+ context window that is great for some really well written articles.

Click to expand...

Yep exactly, it's all about being detailed, giving as much context as possible. Test test test test. Fine tune every time he returns and article, read it all, see what you would like changed or improved, fine tune prompt to achieve that, repeat infinitely until you get what you consider to be your perfect article.

One of the things I'm actually struggling with is making good H2's that are still relevant to the article while also trying to include semantic words/related terms in them. The more I try to force it the less "natural" it feels in terms of article flow, so that's something I'm trying to improve right now.

Another issue is that it's hard to tell how well an article is doing regarding using semantic words/related terms overall per article since we really don't know what's good and what isn't as we just don't have a realistic way to measure it.

I'm working towards a V8 version of my program now.

The way I am handling the LSI and keyword density is by using firecrawl to load a competitors page contents then using pytourch sentence transforms to get keywords. That's only one little piece of what I'm doing to create an article.

chrischris15 said:

I'm working towards a V8 version of my program now.

The way I am handling the LSI and keyword density is by using firecrawl to load a competitors page contents then using pytourch sentence transforms to get keywords. That's only one little piece of what I'm doing to create an article.

Click to expand...

I used to do that a few years back and it was alright and managed to get AI articles ranking that way BUT it wasn't good enough because at the end of the day I was doing and targetting exactly what ym competitors are targetting, I always try to surpass them and provide more value so google has a reason to rank me higher.

Nowadays AI can also do all of that without using python, transformers, etc. Can use AI Agents for it or literally just tell claude/chatgpt to do competitor research for example and they will do the same by visiting the competitor websites

spectrejoe said:

Yep exactly, it's all about being detailed, giving as much context as possible. Test test test test. Fine tune every time he returns and article, read it all, see what you would like changed or improved, fine tune prompt to achieve that, repeat infinitely until you get what you consider to be your perfect article.

One of the things I'm actually struggling with is making good H2's that are still relevant to the article while also trying to include semantic words/related terms in them. The more I try to force it the less "natural" it feels in terms of article flow, so that's something I'm trying to improve right now.

Another issue is that it's hard to tell how well an article is doing regarding using semantic words/related terms overall per article since we really don't know what's good and what isn't as we just don't have a realistic way to measure it.

Click to expand...

So how do you go about it(not asking for exact details). I myself jist have a single prompt and tbh the article sucks and using opeani 4.1 mini.

Do you like feed it competitor websites first or how foes thay work, trying to understand how to go arlund the multiple api calls

DanDD said:

So how do you go about it(not asking for exact details). I myself jist have a single prompt and tbh the article sucks and using opeani 4.1 mini.

Do you like feed it competitor websites first or how foes thay work, trying to understand how to go arlund the multiple api calls

Click to expand...

Well, the way I do it right now I don't micromanage it at all by feeding competitor URLs but I used to do that before the AI could actually browse the web. I'm using ChatGPT Plus with 5.4 Thinking right now (I always use thinking I think it gives better results) but it also works with AI Agents and Claude Opus 4.6.

In a very very simplified way, first began by asking the AI to read a few articles I hand picked from some great websites, backlincko, etc, any authroity website. Then I told it I want the AI to be able to replicate the writing style and to create me a prompt for that.

Then from that prompt I just kept improving it.

I ask it to do competitor research for the keyword, generate semantic search / related terms, etc... Then ocne he retrieves that data for me I tell it to create the article for me keeping in mind those things he retrieved for me.

This is a very simplified way of explaining what I began with. I actualy use ~4-7 prompts per article but 90% of it can be done with the 2 main prompts.

spectrejoe said:

Well, the way I do it right now I don't micromanage it at all by feeding competitor URLs but I used to do that before the AI could actually browse the web. I'm using ChatGPT Plus with 5.4 Thinking right now (I always use thinking I think it gives better results) but it also works with AI Agents and Claude Opus 4.6.

In a very very simplified way, first began by asking the AI to read a few articles I hand picked from some great websites, backlincko, etc, any authroity website. Then I told it I want the AI to be able to replicate the writing style and to create me a prompt for that.

Then from that prompt I just kept improving it.

I ask it to do competitor research for the keyword, generate semantic search / related terms, etc... Then ocne he retrieves that data for me I tell it to create the article for me keeping in mind those things he retrieved for me.

This is a very simplified way of explaining what I began with. I actualy use ~4-7 prompts per article but 90% of it can be done with the 2 main prompts.

Click to expand...

Thank you, got the idea. How much does all these api calls cost per article?

DanDD said:

Thank you, got the idea. How much does all these api calls cost per article?

Click to expand...

I have no clue, I made my own solution using selenium to access chatgpt on browser and pull the data that way / input the prompts that way because I don't think I've ever reached any limits in ChatGPT Browser. I do have ChatGPT Plus but yea.

With API it should be significantly faster and easier to setup but like you said there's a cost per article rather than a simple monthly subscription with fixed price.
You can actually google "CatGPT Github" as a starting step if u wanna replicate it, it's similar to what im doing although you may need to edit stuff to fit ur needs + no guarantee it's updated.

Lincon14 said:

Sounds like a solid comeback plan, especially if you already have years of IM experience and a workflow to scale content quickly. The only thing I’d watch is quality vs volume once you start pushing 10–20 articles a day, because topical authority can grow fast but it still needs decent keyword selection and internal linking to really move the needle.

Click to expand...

I actually decided to go with 3 articles a day to reduce the odds of getting hit with velocity penalties down the line. Maybe later on ill move up to 5-10 a day