[Journey] 1 million UVs/month in 12 months using AI generated content. Let's do it!

[Archemike]

API requests is preferable. How do you deal with their Cloudflare protection?

roger that, so you can use certain parameters in browsers to get around anti DDOS networks such as cloudflare. I've used these in other scripts before for unrelated sort of network bots. there's other methods too.

 

[castoro]

@Sartre thank you for sharing your journey, it has lots of inspiring points but god, it really makes me jealous!

I'm using AI to produce paragraphs for my sites and then I edit them a little, manually inserting images, links etc.
Not going to lie, knowing that you produce 100 same (or better) quality articles in a single day and on autopilot is a little disheartening.
How long have you been doing internet marketing/SEO?
Sadly I've only started seriously for 1 year, but still have a day job so I cannot put all my efforts in this.

I would like to show you the strategy that I'm using (just started) that involves AI but 0 automation, I would like some advices to improve or general opinions, but I do not want to hijack this thread with my stuff and I do not want to bother you without asking permission first. If it is not a problem for you, let me know.

Good luck with your great journey, will surely follow this!

 

[Sartre]

roger that, so you can use certain parameters in browsers to get around anti DDOS networks such as cloudflare. I've used these in other scripts before for unrelated sort of network bots. there's other methods too.

awesome. hit me up in DM if you want to talk more too or just share here. I've been using cloudscraper (https://github.com/VeNoMouS/cloudscraper) for scraping websites that are behind cloudflare and that worked well

@Sartre thank you for sharing your journey, it has lots of inspiring points but god, it really makes me jealous!

I'm using AI to produce paragraphs for my sites and then I edit them a little, manually inserting images, links etc.
Not going to lie, knowing that you produce 100 same (or better) quality articles in a single day and on autopilot is a little disheartening.
How long have you been doing internet marketing/SEO?
Sadly I've only started seriously for 1 year, but still have a day job so I cannot put all my efforts in this.

I would like to show you the strategy that I'm using (just started) that involves AI but 0 automation, I would like some advices to improve or general opinions, but I do not want to hijack this thread with my stuff and I do not want to bother you without asking permission first. If it is not a problem for you, let me know.

Good luck with your great journey, will surely follow this!

go ahead anyone can post anything here I'm not selling anything lol.

I've been doing SEO and IM for 15+ years. I had a quite interesting journey. Started in porn(back when keyword stuffing and spamming comment backlinks worked), then some CPA, got into programming (Ruby), then worked in real SEO for multi-billion dollar companies and got into Python, back to my own biz, started a bunch of solid sites.

 

[spectrejoe]

Thanks for this journey so far! Inspiring really.

I think I got my program almost finished, quality is pretty decent, produces about 500 to 2000 words per article and little to no duplicate content, however, still facing 2 issues:
1- Improving (somehow?) the accuracy of the summarizer I'm using, it produces REALLY good summarization 90-95% of the time, but the other 5-10% it produces nonsense in a sentence or two.

2- Current paraphraser sometimes just duplicates the phrase when they're small:
"Lift some weights and take creatine. Drink Protein Shakes. Drink protein shakes. Eat food high in protein such as chicken breasts."
I guess I can fix this by increasing the minimum length of a phrase to be paraphrased but still scared of getting caught with duplicate content penalties since a lot of the time there are a bunch of small phrases in an article and they get picked up by Copyscape.

Other than that, I think I just need to find a way to scale this, currently, I can create 200-300 posts per day, limited by the time it takes to summarize & paraphrase each sentence, I guess buying a new PC with a powerful GPU would help but that's only if the project ends up getting massive returns.

Also, I somehow need to try and find low competition keywords, finding them is easy, the hard part is automating checks to see if they really are low competition or not in a way that doesn't take too long when going through 20k+ keywords.

 

[castoro]

awesome. hit me up in DM if you want to talk more too or just share here. I've been using cloudscraper (https://github.com/VeNoMouS/cloudscraper) for scraping websites that are behind cloudflare and that worked well

go ahead anyone can post anything here I'm not selling anything lol.

I've been doing SEO and IM for 15+ years. I had a quite interesting journey. Started in porn(back when keyword stuffing and spamming comment backlinks worked), then some CPA, got into programming (Ruby), then worked in real SEO for multi-billion dollar companies and got into Python, back to my own biz, started a bunch of solid sites.

thank you very much, I've sent you a PM.

 

[OccultKingdoms]

Other than that, I think I just need to find a way to scale this, currently, I can create 200-300 posts per day, limited by the time it takes to summarize & paraphrase each sentence, I guess buying a new PC with a powerful GPU would help but that's only if the project ends up getting massive returns.

Maybe buy good auction domain for 2k and rent gpu. That's another way to go about it. Chances of succeeding with something to build on increase a lot. Yes, you need a habit of spending your capital on things like domains and computational power, but that's what seo is about since 20 years.

 

[Sartre]

Maybe buy good auction domain for 2k and rent gpu. That's another way to go about it. Chances of succeeding with something to build on increase a lot. Yes, you need a habit of spending your capital on things like domains and computational power, but that's what seo is about since 20 years.

this is a good tip. I usually use expired domains with some decent trust flow in the same niche. Also buy aged/indexed domains sometimes for important projects.

 

[13385]

Hi there!

How are you dealing with copyrighted words in the text you are scraping? When I was at my peak with splogging I was getting multiple complaints everyday.
A lot of large corporations monitor google and file complaints to your hosting/cloudflare if you use their brandnames. Was a real pain in the ass dealing with the complaints for us.

 

[Sartre]

Hi there!

How are you dealing with copyrighted words in the text you are scraping? When I was at my peak with splogging I was getting multiple complaints everyday.
A lot of large corporations monitor google and file complaints to your hosting/cloudflare if you use their brandnames. Was a real pain in the ass dealing with the complaints for us.

So far didn't have problems. We try to identify own names using an algorithm and look through the articles to tweak the algorithm in general to black list the names manually.

 

[Alma]

Thanks for this journey so far! Inspiring really.

I think I got my program almost finished, quality is pretty decent, produces about 500 to 2000 words per article and little to no duplicate content, however, still facing 2 issues:
1- Improving (somehow?) the accuracy of the summarizer I'm using, it produces REALLY good summarization 90-95% of the time, but the other 5-10% it produces nonsense in a sentence or two.

2- Current paraphraser sometimes just duplicates the phrase when they're small:
"Lift some weights and take creatine. Drink Protein Shakes. Drink protein shakes. Eat food high in protein such as chicken breasts."
I guess I can fix this by increasing the minimum length of a phrase to be paraphrased but still scared of getting caught with duplicate content penalties since a lot of the time there are a bunch of small phrases in an article and they get picked up by Copyscape.

Other than that, I think I just need to find a way to scale this, currently, I can create 200-300 posts per day, limited by the time it takes to summarize & paraphrase each sentence, I guess buying a new PC with a powerful GPU would help but that's only if the project ends up getting massive returns.

Also, I somehow need to try and find low competition keywords, finding them is easy, the hard part is automating checks to see if they really are low competition or not in a way that doesn't take too long when going through 20k+ keywords.

duplicate content doesn't matter

 

[haykuroo]

roger that, so you can use certain parameters in browsers to get around anti DDOS networks such as cloudflare. I've used these in other scripts before for unrelated sort of network bots. there's other methods too.

Are you talking about Cloudflare detecting the headless parameter in the user agent string?

 

[Archemike]

Here's a related article from google: https://intoli.com/blog/making-chrome-headless-undetectable/

But yes there are ways to essentially make yourself look like anything. of course ideally the goal is to look like a one off normal user on a site, and scale that up X 100***** definitely don't want any headless or bot like markings in any http requests or any sort of data.

 

[Sartre]

Here's a related article from google: https://intoli.com/blog/making-chrome-headless-undetectable/

But yes there are ways to essentially make yourself look like anything. of course ideally the goal is to look like a one off normal user on a site, and scale that up X 100***** definitely don't want any headless or bot like markings in any http requests or any sort of data.

I mean making requests look 100% like chrome. I had a problem accessing Q***lbot's API because of CF's auth detecting that I'm not a real browser. I forged the HAR requests pretty well I think.

 

[Alexion]

Thanks for this journey so far! Inspiring really.

I think I got my program almost finished, quality is pretty decent, produces about 500 to 2000 words per article and little to no duplicate content, however, still facing 2 issues:
1- Improving (somehow?) the accuracy of the summarizer I'm using, it produces REALLY good summarization 90-95% of the time, but the other 5-10% it produces nonsense in a sentence or two.

2- Current paraphraser sometimes just duplicates the phrase when they're small:
"Lift some weights and take creatine. Drink Protein Shakes. Drink protein shakes. Eat food high in protein such as chicken breasts."
I guess I can fix this by increasing the minimum length of a phrase to be paraphrased but still scared of getting caught with duplicate content penalties since a lot of the time there are a bunch of small phrases in an article and they get picked up by Copyscape.

Other than that, I think I just need to find a way to scale this, currently, I can create 200-300 posts per day, limited by the time it takes to summarize & paraphrase each sentence, I guess buying a new PC with a powerful GPU would help but that's only if the project ends up getting massive returns.

Also, I somehow need to try and find low competition keywords, finding them is easy, the hard part is automating checks to see if they really are low competition or not in a way that doesn't take too long when going through 20k+ keywords.

You worry too much about copyscape.

Have you seen askinglot, cementanswers, midogguide, heck, even healthline?

Healthline paragraphs are SO generic, they have been used hundreds of times BEFORE them, yet they do fine.

It's impossible for google, as a warehouse, to store and compare each and every article to what's being posted daily. And even if they would, the hashes they would create would still be unique due to a number of reasons.

Imagine quotes, and how many times are being used.

 

[mdevo]

You worry too much about copyscape.

Have you seen askinglot, cementanswers, midogguide, heck, even healthline?

Healthline paragraphs are SO generic, they have been used hundreds of times BEFORE them, yet they do fine.

It's impossible for google, as a warehouse, to store and compare each and every article to what's being posted daily. And even if they would, the hashes they would create would still be unique due to a number of reasons.

Imagine quotes, and how many times are being used.

Your post is accurate, but note that healthline seem to be authority domain in Google's eyes. Other than that you are 100% right tho.
News, product descriptions, product specifications, ebook descriptions are being used over and over again and CAN'T be matched as duplicate content because there is no way making it non-duplicate.
The thing I believe in is that Google analyzes whole article as one entity, not every paragraph or sentence separately (that's why most sites doing similar things as those you mentioned are already indexed and ranking for over a year).

It requires too much computing power and moreover they don't give a single fuck about duplicate content as long as it matches search intent.

 

[spectrejoe]

You worry too much about copyscape.

Have you seen askinglot, cementanswers, midogguide, heck, even healthline?

Healthline paragraphs are SO generic, they have been used hundreds of times BEFORE them, yet they do fine.

It's impossible for google, as a warehouse, to store and compare each and every article to what's being posted daily. And even if they would, the hashes they would create would still be unique due to a number of reasons.

Imagine quotes, and how many times are being used.

I think most of these sites spin / paraphrase the text at least, otherwise how come all the autoblogs copying content through RSS feeds and stuff with a billion posts targeting low competition keywords don't rank at all?

 

[Alexion]

I think most of these sites spin / paraphrase the text at least, otherwise how come all the autoblogs copying content through RSS feeds and stuff with a billion posts targeting low competition keywords don't rank at all?

They don't. It's the exact same answer from Google snippet. But they copy one paragraph from one, site, one from another, and the final article, as a whole, is unique. I can bet my $0,83 that they don't auto check beyond that.

You said a sentence wasn't passing copyscrape. I was addressing that issue, that you don't have to worry for one sentence. Even paragraphs are ok.

It's just my opinion, don't take my word for it. But there are plenty of examples ranking on page 1 with plagiarized content(even without knowing in the case of healthline).

 

[spectrejoe]

They don't. It's the exact same answer from Google snippet. But they copy one paragraph from one, site, one from another, and the final article, as a whole, is unique. I can bet my $0,83 that they don't auto check beyond that.

You said a sentence wasn't passing copyscrape. I was addressing that issue, that you don't have to worry for one sentence. Even paragraphs are ok.

It's just my opinion, don't take my word for it. But there are plenty of examples ranking on page 1 with plagiarized content(even without knowing in the case of healthline).

Interesting, I'll see, got a few AI websites going on. I have a 1 year old website with 24k posts but it gets like 30 clicks from Google and 700+ daily from Bing, assumed it was just Google not wanting to rank it due to duplicate content

 

[Sartre]

Interesting, I'll see, got a few AI websites going on. I have a 1 year old website with 24k posts but it gets like 30 clicks from Google and 700+ daily from Bing, assumed it was just Google not wanting to rank it due to duplicate content

successful sites got strong backlinks

 

[spectrejoe]

successful sites got strong backlinks

Right, that's the thing though, I can get 500-1000 visitors a day from Google using unique content and with 100-200 articles on "whitehat sites" with no backlinks but I can't even get more than 30 clicks a day from Google on a 24k posts website? There must be something I'm missing, especially since the 24k posts site actually got quite some decent backlinks over the year. Which is why I think that Google definitely factors in content uniqueness