VesperSEO

This project was initiated by myself and my friend Timothy Booker, both of us having a background in online gambling affiliate marketing.

1. Creating a single website that boasts multiple language versions in a current SEO landscape is not feasible. There was a time when multisites used to get an additional boost in the search engine rankings, but that hasn’t been the case in several years. Also, I think diversifying sites is a good strategy to avoid your site getting hammered by spam attacks or by algorithm updates.
2. Instead, there should be a scalable way of launching new sites without needing more human resources.
3. WordPress is not an optimal solution when operating multiple websites. This results in a constant need for updates.
4. The system should have elements of AI incorporated into it to assist the webmasters.
5. Affiliate offers should be controllable in the CMS directly.
6. Data related to the affiliate products or services should be structured and stored so it can be shared across different websites, despite the language or target market.

We decided to start the project by creating crypto affiliate sites targeting nordic markets. The niche and the target market were medium to low competition. We created 10 small and niche (within niche) affiliate sites targeting different segments in the crypto space.

We started to see some initial traction, but shortly after FTX collapsed and the volumes went down significantly.

Two months later ChatGPT was rolled out. We started thinking about bringing GPT content creation capabilities into the system. We didn’t want to use ChatGPT directly as the content is not very accurate and has bias baked into it.

Instead we decided to leverage different data points together with GPT API to create content based on data. The process was not easy. There was a lot of trial and error with the prompting, but we were able to create pretty good content for search engine ranking purposes. The best part? It was a fraction of the cost and the time required compared to using human writers.

Early in the process, we created a site called slotindex.org. The site has over 1000 free to play slot games. Everything on the site is Ai generated content. It cost 400$ to generate the whole site. The content was generated using GPT-3.5. This was before the 4.0 came out. The cost would probably be even less with the current token pricing.

The first SEO question would be, does it work?

It depends on how you define “does it work”.

Did it get indexed? Yes, but not everything in the site get indexed.

Does it rank well? No, but the site hasn’t been pushed with links or hasn’t been fully honed to the ideal ranking level.

After getting the Slot Index online we went on a publishing spree. We launched roughly 30 sites in two months. The sites we launched were targeting different niches and multiple different markets. We didn’t hire external help. We just wanted to see how much we could push the system with minimal human resources.

Here comes the sad part:

It’s a cool product that does cool things. However, cool products don’t necessarily translate into a business.

We tried monetizing the system in various ways, but nothing really materialised. Around November 2023 we were both getting exhausted. We haven’t seen any monetary results even though we’ve worked so hard. We decided to put the project on pause for the time being. Maybe we come back to it, maybe we won’t.

I learned tons from this project. I learned that when it comes to SEO content creation with AI it’s not black and white. It also takes a lot of time to see if something “works” or doesn’t work.

Out of the box ChatGPT content is not the best content for SEO. The strength of ChatGPT lies in editing, improving and brainstorming. If you can get AI to write your content from data, it can do a much better job than your average freelance writer.