A couple of months ago I was toying around with a couple of AI content tools. I know the SEO community can be pretty divided here but I wanted to at least see how far our machine overlords had come.
I took an established site in a niche I normally pay $50 for an article. Or at least an hour of my own writing in a niche I have zero interest in writing for.
Within 20 minutes I had an article ready. It was better quality than most of the freelance writers I’ve worked with and after a couple of weeks, it was one of the top trafficked pages on the site. To make sure it wasn’t just an odd fluke, I published another two dozen articles on the same site. Each one took less than half an hour and they all pull in consistent traffic.
So far so good with one site, how far can we push it?
A couple of weeks ago I started a couple of spiders up hunting for high traffic domains with low domain authority. The project was an eye-opener and I’ll probably do a case study on chasing these niches at some point in the near future.
What it did in the short term was really drove my demand for content.
My strategy is to start a large stack of websites at once, fill it with a ton of content and then come back to scale the easy wins further. Normally, this is an expensive prospect. It’s been working for me but capital and editing time has been the limiting factor.
I have a hard time outsourcing everything. It’s an ongoing issue and I just haven’t found someone I could trust with full autonomy on quality control yet. So editing AI generated content isn’t going to be that much of a jump from editing freelance writer content.
I can’t let the AI handle content itself. I’m well aware there’s a trend for this at the moment but it’s just not part of my strategy. I find if I edit the content myself I’m getting decent quality which is at least on par with the quality I was getting from most freelancers. Often better.
It’s a tool to help writing – not a replacement for good writing.
So I did the math.
If I take a high traffic and low DR competitor and get a domain – that’s $10. Call it maybe $30 including hosting.
If I want 100 posts to test it out that would normally be about $4,500. If I can use AI content to keep it under 10 minutes then the worst case that’s 1,000 minutes or 16.6 hours. Let’s round that up to 20, include some time for a coffee break and I can still produce a new website a week.
I’m chasing low DR sites getting 30-50,000 hits a month so even a low RPM display site would get a couple of hundred bucks a month. Let’s lowball that again and say the site makes $250 a month. With a 30x multiple that’s $7,500 for 20 hours of work.
Even if I spend $1-2,000 on outsourcing some outreach to beat the low DR competitors that’s still a good return.
Obviously – these numbers are pie in the sky. Some sites will flop and some will do better. Some will take longer than 20 hours and some will take less. A similar model is how I made some of my easiest wins in SEO.
To improve my chances (and keep my editing time sub-20 minutes) I added a couple of extra hoops.
- I’m chasing niches where the content is fairly systematic or easy to write about. It can’t rely on in-depth research because the AI will just produce nonsense.
- I wrote a script to handle the keyword queue, generate images and draft the content on WordPress for me. Cutting down the admin as much as possible makes it quicker and means I can focus just on editing.
- Ideally, I didn’t want to rely on display ads. Personal preference here to make a bigger RPM but also if I can get people clicking through to a landing page then the bounce time of the site is improved.
So, with this all in mind, I got started.
Site 1: Immediate Results
With a plan in place, I started the first site.
I got a bit lucky with this one. I found a low-competition niche with a ton of underserved longtails and decent affiliate options. I’ve worked with a semi-related topic so the content was easy to edit.
I don’t put a whole lot of stock in the general SEO advice you sometimes hear. I’ve heard some crazy rules like don’t post more than 10 posts in your first month or you’ll look unnatural!
There’s definitely evidence that content velocity and freshness matter. Keeping content updated and a site fed with new articles drives good growth. If I know I’m letting a site sit for a while I’ll schedule posts out over time but if I’m actively publishing – I want those pages in the index ASAP.
Within the first week, this site was generating double-digit traffic and even had affiliate hops in the first few days.
I haven’t set up Google Analytics on this site yet but this might be some of the most solid trending growth I’ve ever seen on a new site. As it was the first site, it took a little time to get my workflow going smoothly and some of the earlier posts were a little rough around the edges but at this rate, that site is headed for 1,000,000 published words over the next couple of months.
I’ll probably slow down the publishing a bit as we start new sites up but I do think this one will become an easy winner. If that growth continues it’ll probably be some of the first affiliate income from the project.
Site 2: AI vs AI
Site 2 was a first for me.
I’m no stranger to aged domains (in fact, I’ve just started another case study bullying a DR 0 with an aged domain) but this was the first time I accidentally registered one that used to exist.
With the hundreds of domains I’ve owned over the last couple of years, it probably should have happened sooner.
I know some SEOs advise checking for this sort of thing but I rarely bother. The days of EMDs are behind us and I tend to use brandable names so I was surprised to find this site already had a handful of backlinks and was getting traffic to an old page.
I checked the history and it looks clean from what I can see at a glance. Looks like it was a hobbyist blogger and the handful of links it has won’t carry much weight. It was parked for a while but it looks like the Google index is still using some old meta descriptions for it.
I set up the site with some basic branding and the first post to get the ball rolling.
This site is also in an interesting niche. The spider flagged a DR 2 with almost 50,000 estimated monthly hits and growing and when I checked into the site it was either entirely AI-driven or a pretty bad freelance writer. It’s word salad.
This is not my favorite niche in the world. It’s going to be pretty low RPM on display ads and affiliate options might be a hard push. It could maybe be a decent lead-gen for the broader niche but I’ll put 20 hours into the content editing and see where we wind up.
We should be able to get this site to around 80 posts and we’ll check back in a while to see how it’s doing. Even if we take half that traffic and stick on display ads we could probably flip the site to someone who wanted to push it further.
The Case Study
This is an SEO case study which means it’s going to take a while. Site 1 has had some absurd growth to start with but that won’t be the norm.
I do like to start several sites at once so they can start getting indexed sooner rather than later. I’ll get baseline of content published to these two and probably open another one or two within a month.
AI vs Outsourcing
I put tens of thousands into content outsourcing last year with… mixed results. I don’t like the market rate of paying per word because you’re just incentivizing waffle content and the content site market is full of articles that would make your eyes bleed.
AI by itself is even worse. I’ve seen it go off on strange tangents insulting the reader and threatening them with death. Probably not great for conversion rates.
I personally think a good writer who is practiced with AI content tools is a lethal combination and if this case study goes well my next goal will be finding someone who can handle that on an hourly basis.