SEO hacking with AI and why most established companies are not doing it

And Marketing BS is back!

Thank you for your patience in November. I knew it was going to be a crazy month. I paused paid subscriptions, but I still hoped to get a handful of posts out to keep up the momentum. No such luck. My family has some more extended travel over the Christas holidays, but I hope to get some extra posts pre-written before that travel begins. Now on with the show!

---

This Week’s Sponsor

Get this playbook launched by Amplitude

Keep acquisition costs low and maximize your return on advertising and marketing spending. Take a page (or two!) from our Acquisition Playbook and crush the competition. Learn: Acquisition fundamentals, How to lower CAC and maximize ROAS Tips from industry experts. Learn more here.

---

Traditionally SEO work is divided into three types:

1. White Hat

2. Black Hat

3. Grey Hat

White hat means doing the SEO stuff that Google wants you to do. Following Google guidelines to make your site fast, informative, and easy for the Google robots to read and understand. In so doing you should organically get links from other white hat websites, which Google will take into account when it decides where to put your site in the rankings.

Black hat means trying to reverse engineer Google’s algorithms to rank higher than your site “should” given the fundamentals that Google thinks are important.

Grey hat falls somewhere in-between.

Google has always been at war with Black Hat SEOs. Who is winning at any given time shifts. Sometimes Black Hat SEOs develop a technique that “tricks” Google’s algorithms and allows a site to rank very high very fast. But Black Hat SEOs are on a constant treadmill battle against Google itself. Every thing they do to improve their sites ranking is by definition temporary as Google tries to tweek their algorithm to eliminate the arbitrage opportunity.

There have been times when Black Hat SEO worked for an extended period, but eventually Google’s technology allowed the company to eliminate the effectiveness of the technique. The most glaring example was in 2007 when many established companies were “hit by the Panda update” and had their SEO traffic cut by 50-90%. I have not heard stories like that recently, but I think that is for two reasons:

1. Google has been winning the war against Black Hat SEO. There are fewer opportunities for companies to build significant SEO traffic using "disapproved techniques”, so therefore fewer opportunities for that traffic to evaporate upon an update

2. Big companies learned from 2007. If you build a business on Black Hat SEO, and then get hit by an algorithm change it can destroy a company. Companies build their cost structure around their revenue, and if your company has used Black Hat SEO to generate $30MM in annual revenue, and then 90% of that revenue disappears over a weekend, the rest of the business is put in jeopardy. Better to grow slower but in a more sustainable way

So that was the state of the world about a year ago. But now we have AI and LLMs

Using AI to Accelerate SEO

When AI was picking up steam many people were worried that Google would penalize AI generated content. But two things happened:

1. AI got so good that there is no way to tell when content is AI generated (A few weeks ago a study was published that demonstrated that AI generated human faces are now MORE likely than actual human faces to be identified as real)

2. Google itself came out and said that AI generated content was fine — it was the quality of the content that mattered, not how it was generated

At the time I pushed readers, and portfolio companies, to dramatically accelerate content generation using AI. There was no reason to employ someone to write a few blog posts per week. It was just a make-work project. Companies should figure out what ALL the content they should be generating is, and then publishing a new piece every hour until they have created everything that they could possibly need. If that is too much for their website, they should be creating flanker websites to create the content. Content generation is now (effectively) free. That should change the math on how a business is run.

And yet clearly that is not happening. At least not with established companies. There ARE some smaller companies that are taking advantage. These are the companies that in the past would be doing things called “growth hacking” and Black Hat SEO. Only AI generated content is NOT Black Hat. At least right now.

Here is a great example I came across on Twitter a couple of weeks ago that is now getting some buzz (screenshot of first post in the thread, then quoted as text):

Full thread (bolded are my choices for clarity)

We pulled off an SEO heist using AI.

1. Exported a competitor’s sitemap

2. Turned their list of URLs into article titles

3. Created 1,800 articles from those titles at scale using AI

18 months later, we have stolen: - 3.6M total traffic - 490K monthly traffic

How we did it? 18 months ago, this was a totally manual process. Today, there are tools to help automate 90% of the process. Here’s how:

1. Find a competitor Which website ranks for the keywords you want to rank for?

• Brainstorm

• Use Ahrefs/Semrush

• Manually search on Google

2. See their sitemap To find your competitor’s sitemap, add the following to their domain: http://website.com/sitemap.xml

Their sitemap shows a list of all the URLs on their website. However, the URLs in the sitemap must be descriptive to get the best results... For example:

http://website.com/t/1043423

http://website.com/how-to-make-soup

3. Export sitemap Google “convert XML to CSV” and use one of the free tools to export the sitemap as a CSV file.

4. Generate ideas Using the list of competitor URLs, it’s time to generate article ideas. Add the URLs from the sitemap into Byword’s Write From URLs feature and scrape the URLs.

Byword generates article titles and outlines based on the competitor URLs provided. For example: http://website.com/how-to-make-soup… might get translated to: A Beginner's Guide to Making Soup

5. Write with AI Optional: Edit the suggested titles and headings for the articles...

6. Publish quickly Publish the content straight onto your website using CMS integrations (to save time).

Results? Using this process, we published 1,800 articles in a few hours. Since then: - 490K monthly traffic - 3.6M total since publishing - 13K keywords on page 1 of Google The best part? We’ve taken that traffic from our competitors.

The author was pretty proud of himself, but the general reaction has been something approaching disgust. “How could he do such a thing? This is clearly bad for users!” It didn’t help that he framed it as “stealing traffic from competitors”. It just felt mean.

But is it really bad for consumers? It would be if the quality was low. But its unclear the quality in this case IS low. It certainly doesn’t have to be low. AI can write better than 90% of people right now. Are you so sure the writers at your company are in the top 10%? I can guarantee the average piece of content on the web is not written by a top-10% writer.

Where AI writing is bad it is bad when it is “SEO-junk”. I came across some the other day when looking for a sleep-away camp for my chess-loving daughter. I found a camp website that talked about “chess sleepaway camp”. There was a long articles about all the benefits of teaching chess and camp combined with other activities. 70% down in the article it mentioned that their camp has “weekly chess games”. Really? It’s a chess sleepaway camp that plays one game of chess per week? I did some digging and this same camp has articles on just about every potential camp activity and why that activity makes a great camp, and how focused their camp is on it. It was all SEO spam trying to drive long-tail clicks (and clearly no one stopped and thought — “is anyone going to convert off this page talking about our weekly chess games targeting people looking for a chess camp?”).

But the technique Jake is using is NOT SEO spam. He is trying to duplicate the account structure of a competitor and create content that is as good or better than what his competitor created. It is only spam to the extent that his competitor has already created (successful) spam.

The author is promoting his AI tool “Byward.ai”, but the general process should not be very difficult even without a tool like that. I would make a few changes to the process, like instead of dumping 1,800 articles in a few hours I would spread out the publication and drop one every 30-60 minutes. It would mean delaying things a little. 1,800 articles dropped every 30 minutes would take a little more than a month to go live, but I think the trade off should be fine. It would allow the Google robot to index the pages over time and see an active website, rather than a site dumping a bunch of content one-and-done. I am not convinced my way would be better, but when it comes to this stuff I am generally more conservative.

The main point is, is that most companies are not doing it his way OR my way. Most companies are still publishing 2-3 articles per week, because that is what they have always done.

AI is going to radically change the way we do business, but like all new technologies, the opportunities created by the technologies happen much faster than the processes followed by individuals and companies. The current people running SEO, at least at big organizations, generally don’t have the temperament to make a radical change like Jake is suggesting.

What happens when everyone is doing it? It’s possible Google will shit it down. There are a few ways they could do it. They could increase the value of the first company to create an account structure and then penalize anyone else who does the same. This would encourage innovation, but discourage everyone else from “catching up”. Google could decrease the value of content in their search results. This will happen naturally as more answers get picked up with “zero click”, but otherwise I don’t see content as not being important for content-related searches.

I would love to hear from my readers though on whether ANYONE has started doing this at their company. Happy to keep the results confidential.

I hope everyone had a great holiday. I just un-paused premium subscriptions and members will get their second post this week on Thursday or Friday. If you haven’t subscribed already, you can do that here:

Keep is simple,

Edward

P.S., This week’s image was built by plugging in the essay into ChatGPT and asking it to create a meme. The details weren’t great but the basic idea of an AI saying “hold my beer” made me laugh. It also failed to make the actual image. I changed the details to what I wanted and then iterated on the image until I got something I liked. Then, while the AI can create text now, it was not doing what I wanted it to do, so eventually I took the images generated and edited in the text on my own with an image/text editor. I am pretty happy with the final result! Maybe I will do a meme every essay now!