Spun spam sites earnings went up 3x

[Philip J. Fry]

I found an interesting trend that might be interesting for the people of this forum. About 5 years ago, I created a site using spintext and a list of nouns. It's a site where each page is "____ $Noun ____" and the content on the pages are all spun. An example is "Is it hot in Sturgis, ND?" Then, when you land on the site, you can vote in a pole and, once you vote, you'll see what the results are. There's forty thousand pages on this site for Nouns. The site doesn't pass PageSpeed Insights. I haven't added AMP to it or done any work since that one weekend I worked on it years ago.

Guess what? Revenue went up from $50-$100/month for the past few years to $250/month for the past few months. All this is telling me is that, for these obscure long tail queries where there's almost no other sites targeting the queries, Google doesn't have many options to return, which makes piece of shit, spam, spintext sites still rank and earn.

It was an expired domain with several thousand spintext pages on it. The key is that it can't be a query where, if you type in the query, Google will give you the result. An example of this is "What time is it in Sturgis, ND?" You need a query where people would need to read a webpage for this site to be replicated again, for another topic.

Good luck.

 

[bernard]

I'm not sure if you can say anything from this example, but there is something about moving from semantic analysis to contextual analysis, using more and more AI.

We're entering into an age of AI written content, so word for word, might not be that important as it used to be, the more important is does the returned answer the intent.

 

[Boy]

The only thing that irritates me from your post is that Sturgis is in South Dakota, not North Dakota. Unless, of course, you're literally targeting people that search for incorrect stuff like that.

 

[Philip J. Fry]

I was curious and browsed some more. I found another good example but I doubt a spun site can outrank niche.com for these terms. The keyword pattern is "is $UniversityName a good party school?" Then just make spintext about partying, that state, that city, the weather there, alcohol, etc, and spin it for the list of universities you have. There's tons of them. The problem is that you're up against a CF60 domain, Niche.com, and a CF50 domain. Darn!

I'm not sure if you can say anything from this example, but there is something about moving from semantic analysis to contextual analysis, using more and more AI.

We're entering into an age of AI written content, so word for word, might not be that important as it used to be, the more important is does the returned answer the intent.

That might be true. If you make a 40,000 page site about a topic, even with spun content, the whole site would be quite relevant about the topic, especially if you include subparagraphs about topics that are related to the main topic. So, with the party school example, it'll be including subparagraphs about university, higher education, partying, music, dancing, fraternities, sororities, bars, clubs, hookups, weekends, etc. The list of topics that are semantically related is a lot. So, instead of writing content that includes LSI keywords, you just create subparagraphs that are about a topic, where the topic is an LSI keyword. It'll check off the LSI relevancy requirement for the page and domain. It's quite stupid and simple at the same time.

The only thing that irritates me from your post is that Sturgis is in South Dakota, not North Dakota. Unless, of course, you're literally targeting people that search for incorrect stuff like that.

SD