How I'm using AI for SEO

I'm launching Kablio.com, a jobs platform that will need Google traffic, and so I spent a few weeks learning about SEO. What I thought would be boring turned out to be interesting.

Google "organise the world's information and make it universally accessible and useful" (that’s their official mission). SEO is about understanding and gaming their ever-evolving system. So if you're curious about how information flows on the internet and enjoy writing, SEO might be for you. It’s not rocket science either—just look at the backgrounds of SEO consultants out there; it doesn’t take a PhD.

The internet has enough blog posts written by wannatrepreneurs titled “everything I’ve learned about SEO.” (For top learning material, check out the Ahrefs beginner’s SEO course, How the Fxck, and Grow and Convert’s blog). So instead, I’m writing one on how I’m using AI for SEO.

A few points to note

LLMs (currently) just produce mirage content

The best AI models struggle with ideation—they generate mirage content.

Mirage content is content that appears ok at first glance but lacks value, is written only to index and leaves you thinking “no shit”. How often have you clicked on an article from Google to find bland, rehashed, and unoriginal content?

Mirage content has always been an issue, but with the rise of AI, which can produce mirage content at scale, the problem is intensifying, and Google is actively fighting it.

SEO is broken

Soon, LLMs (some of which are being specifically trained to create indexable content), will flood the web with content indistinguishable from the best human-written material. Understanding this shift is super important for managing expectations and preparing for the next phase of the internet.

There still a good SEO opportunity over the next year or maybe even two (I think)

SEO remains effective and will continue to be until LLMs get better at ideation. In fact it’s never been easier to create scalable content that ranks well using AI. However, you need to be strategic, as Google is actively penalising AI-generated content.

Note that it’s obviously important to use all the classic SEO strategies: creating topic clusters, focusing on pain points, content-customer fit, specifity strategy, choosing low KD keywords, creating a unique piece of content for each keyword and all the rest.

How I’m writing non-mirage content at scale

Use AI to write but not to think

AI is good at writing but struggles with generating original ideas. You need to provide specific points for it to format into coherent, keyword-rich content.

My process involves jotting down all the points I want to make, often incorporating ideas from other sources, and then asking AI to format each paragraph. I handle this one paragraph at a time, as the quality of LLM responses decreases with longer outputs. Finally, I ensure the entire article flows smoothly.

Where I get my sources of info from

I’ve launched a jobs platform focused on the construction, energy, and engineering sectors. While I haven’t worked in these sectors, career success in them is about 75% people and management skills. So I bring popular management ideas from the tech sector, learned from experience and classic books such as Working Backwards, Radical Candour, and Principles.

Reddit and Quora are invaluable for finding straightforward insider information. For example, if you’re curious about the value of an engineering certification, professional websites will give you the usual promotional content, but Quora and Reddit will provide unfiltered truth. I often copy and paste insightful community responses into ChatGPT to generate excellent blog paragraphs.

Note that it’s super important to reformat all the ideas you gather, as Google penalises duplicate content.

Topical authority

Topical authority is a crucial SEO strategy where a website becomes the trusted source for specific subjects by extensively covering them. This involves producing comprehensive content about a topic. For example, to establish authority on “protein powder,” a single article isn’t sufficient; you need to cover all aspects of protein powder, creating a network of interlinked articles. This signals to Google that your site is knowledgeable and trustworthy on the subject.

Topical authority can overcome a low Domain Rating (DR), allowing underdog websites to compete effectively by focusing on building comprehensive content first. Also, it accelerates the ranking process (it usually takes around five months for initial content to rank on Google’s first page).

So I’m using AI to produce (good) content at scale that covers certain topics in full. However because this tactic is increasingly easy, Google is increasingly placing more emphasis on EEAT, which is heavily influenced by your quality and quantity of backlinks.

Backlinks to build topical authority

High-quality backlinks from reputable sources within your niche reinforce your site’s credibility and signal to search engines that your content is valuable and trustworthy on a specific topic.

It’s hard to get quality backlinks, especially as you’re getting started. There’s a real chicken and egg problem, and it also means you need to produce content that people want to link to, which means you can’t completely outsource your content production to AI (yet).

Foundational backlinks

Getting your first few backlinks as an underdog is hard because you’re a nobody. However, getting “foundational backlinks” to achieve a Domain Rating (DR) of at least 10-15 is crucial; without this, your content won’t rank, no matter how good it is.

To build these foundational backlinks, post your site on various product directories and launch it on Product Hunt. A lot of these have dofollow links, serving as valuable foundational links. (I know this topic isn’t related to AI, but I feel this advice is not given out enough).

Using Programmatic SEO for Job Searches

Programmatic SEO automates the creation of hundreds or thousands of pages to target a wide range of keywords. By generating specific content for each keyword, you can effectively meet search intent. For example, if someone searches for “Civil engineer jobs London,” programmatic SEO allows you to create a dedicated page that directs users to relevant job listings in London, efficiently satisfying their search needs. AI doesn’t significantly impact this tactic, as it focuses on creating numerous targeted pages rather than producing content at scale.

Conclusion

So that’s how I’m using AI for SEO. Am I producing 10/10 content? Definitely not. But I’m producing pretty good content at scale, which is much better than content with no human-AI involvement.

Let’s see where this goes. I’ll report back in roughly six months and share my learnings.