Hey,

I tweeted a long, rambling tweet last week about how I felt the SEO meta was with AI writers and how to get ahead.

This one’s my expanded thoughts on this.

(This doesn’t apply to some areas, e.g. SaaS content marketing where the product/content is novel and AI can’t do it.)

But… yeah.

The differentiator between AI and human writing now really isn’t quality of writing.

AI writers can output perfect grammar in an instant, and the written voice and quality has improved a lot over the last year.

With the right prompts (an important but learnable skill) you can get AI to output decent level content in basically any style of voice you want.

It can be trained to do first-person, second-person, etc, and write with more life and energy than the standard boring academic prose it gives back without better prompting.

So, what’s your differentiator now?

​

---

AD:

So, you’re an affiliate marketer. You don't do this just for fun... One thing's for sure – you think money. You think revenue. You think traffic boners!

​Odys Global is your source for a guaranteed site launchpad. You get premium aged domains, exclusive mentorship, and access to pretty much everything you might need to start a money-making website. And it’s 24/7.

Think brandable names, clean history, and big SEO value – these are just a few of the things you get when acquiring an Odys domain. Oh, and you’ll save tens and even hundreds of thousands of dollars on link building. How? Simply because these domains are thoroughly vetted and cherry-picked by the best in the biz.

And, speaking of the best in the biz, you can always get a hold of one of the Odys Mentors – industry experts, who are happy to help you out with your site-building strategy, monetization, or anything else you need assistance with.

​
Kickstart your website with domains that aged well – REQUEST ACCESS HERE​

View Odys aged domains here​

---

1. Your research skills
2. Vision: your ability to structure content in the best way – for humans and search engines

Yes, while AI can do most styles of brand voice, it’s still fairly generic on most topics – especially if your niche is more technical.

It prefers to jump to the next topic or section, rather than get really deep into that particular point. And that style of hop-to-the-next-lilypad-content instead of doubling down with actionable info is just not going to be successful in 2023 and beyond.

This is good for you as a writer – AI is still relatively surface-level, unless you create very detailed prompts for specific sections, and tell it which talking points to discuss.
​

1. Research

So, the first way human writers can thrive is by doing the research, and really understanding the topic.

It used to be that research would be perhaps a third, to a half, of the work? (The other part being writing.)

Now it’s more like 80% research, 20% prompting and editing.

Vanilla AI prompts are generic, remember?

But not AI prompts with dynamic, informative data so that it can output info-dense, high-quality content.

Human writers can thrive and stay relevant via these research skills, finding the extra information others don’t have, or at least adding all the available info, prioritising what is important enough to be a focal point of that H2, and ensuring it’s the most comprehensive yet concise answer possible.

You don’t need to just write more (that’s probably bloat and fluff – which AI is good at, but isn’t helpful).

You just need to write dense.

In-house, we call this six-pack content.

It comes from the Comedians in Cars Getting Coffee episode with Hasan Minhaj and Jerry Seinfeld.

Hasan talks about how when he’s first writing the material for a special, and testing it out in smaller comedy clubs, it’ll be good, and there’ll be good material in there, but there will also be a lot of extra fat.

Through refining and performing and editing, the fat gets chopped off, and the material gets leaner and leaner, and more and more packed with the top-tier stuff that gets belly laughs from the crowd.

This is how we try and write content – be super dense with the stuff that matters, and as short as possible otherwise.

AI struggles with this because it doesn’t have as deep knowledge about each topic without more specific prompting.

I mean this from an information and helpfulness perspective… but naturally packing this in densely helps with entities, and entity salience, if you’re structuring your sentences well with the object of the sentence at the beginning.

This is similar to a lot of the great info Koray has on the subject.

Your research allows you to output this info-dense content faster than ever with AI. The actual act of writing, and the writer’s block of structuring these sentences can be bypassed now if you feed your research into AI, and prompt it well.

Good researchers who are bad writers can now outperform good writers who are not disciplined enough to do adequate research.

From there, though, you still need the SEO and vision skills to know what to cut, edit, refine, add, and restructure from this good base – in order to have the best possible piece of content.

It’s also why a number of content agencies, who focused on volume instead of quality, are going to struggle.

As an equity holder in MyContentPal, I speak with Craig Dewart, the founder, often. He understands this well.

Writers, and their ability to output 2000 words consistently every day, are no longer novel or unique.

But these writers can focus on quality, on research, and on the other areas of content, and augment themselves with AI where it can help with no downside in content quality, to offer more value to clients than ever before.

I know Craig understands this, and I think he’ll move towards a model that benefits clients with more value than ever – and I think AI is key to this. I can’t talk about much more here yet.

There’s a Google patent that Bill Slawski breaks down amazingly (RIP Bill), called information gain score, and how Google may use it to rank pages.

In my opinion (not going to get into full detail for my whole theory here, but can do in another newsletter), information gain score as a total nominal value across site, or average IGS value per page, both informs Helpful Content scores to an extent, and if there is a separate on-page EEAT score, then forms into this (as well as entities, etc).

Deeper research to surface these unique insights (or, just be an actual expert… lol) is essential for this reason.

But often, experts, or excellent researchers, still fail.

There’s another key element – your vision and artistry to create amazing page experiences: how you structure this content.

​

2. Content Delivery

So, you’ve sorted info denseness and edited it so it reads well.

The editing forms part of the content delivery – on a micro-level – but you also need to think about it across your entire post from a macro-level.

What I mean is, yes, you can pack in all the detail and create helpful sections in an article, but you also need to structure that article to be as helpful possible.
​

So, things like:

1. Making sure the Main Content is as high as possible in the article (basically the answer to the question, or whether the product is worth buying and why in a product review) – it’s even part of the Page Experience doc Google has.
2. Making that Main Content as digestible as possible – for example, if it’s “how much does a vacuum cleaner cost”, then perhaps a table right at the top after your featured snippet paragraph answer, with the different types of vacuum, and the different price ranges for each.
3. Ordering sections based on importance and relevance for the topic itself – having the info you deliver in the right order to maximise engagement, and deliver the best info first (not just the main answer, but the supporting info beyond this)
4. Covering the topic in the right depth and technical level based on the search term, intent, and audience – if it’s “what is a black hole”, it’s not correct content delivery to go into the complex calculations about quantum hairs and the maths around event horizons. It’s clearly a beginner wanting an easy-to-understand overview – and you’re missing the user intent by not explaining it accessibly simply.
   ​

All these require focus, discipline, and the vision and artistry to create the best page for the user, and for Google.

For Google, this is mostly about the main answer right at the top, but it’s not the only factor in high-quality content.

It’s also about finding more creative ways to be helpful. I gave the example of a supporting table for different price ranges for the “how much does a vacuum cleaner cost” example.

Another example for a “5 common fixes for vacuum cleaner issues” would be to summarise the 5 fixes in numbered bullet points, with each bullet linking to the heading that this covers this fix in more detail, via HTML anchors.

Ordering sections based on relevance, and the right way for people to get info, is also important and requires understanding the reader. AI is actually decent at this though.

But for things like FAQs, the AI writers typically just write FAQs based on the People Also Asked parts for the main keyword.

These are often inaccurate or just completely irrelevant. Delivering content well requires understanding what extra questions people want to know, though you can also prompt AI for finding better FAQs now.

More important than ever is the skill of being a good content deliverer – content is so commoditised now with AI, that how you display and render this is the differentiator, as well as the research.

​

So in my opinion, switch any per-word writers you have to per-hour, per-piece, or just salary them.

​

Incentivise them to focus on research and to enjoy finding the juicy facts and supporting data and figures and stats that nobody else has. Tell them the only priority is creating the absolute best possible article.

Then, save time back for them with good structure and a good brief (AI helps a lot), and allow them to use AI to support and turn their research into prose (this requires a lot of refining and skill – you can’t just give free reign here).

Fact checking is more important than ever – again, this requires discipline and focus, like the research and content deliver parts.

But with these new focuses, you can create far better content is far less time, leaving you with more resources remaining to create amazing infographics, buy links, and to promote the content across social media, email, etc.

In my opinion, having writers just manually research and write is ineffective now (we still do it to an extent though, so I'm a hypocrite here) – and your competitors who are utilising AI and other tactics will outproduce you.

(Again, ignore this if you’re in a super technical niche where AI cannot be trusted to a high degree, or in SaaS for proprietary info and content, or any other area where AI cannot/has not been trained well).

And, pure AI is a terrible idea. None of what I publish would ever be detected as AI. It’s literally not AI. It’s human data collection, human content structure design, and carefully inputted data and points to make, which AI helps structure in a specific voice to save the manual time of typing these sentences up. It basically bypasses the writer’s block.

Pure AI will not work. You have no moat, are making yourself obsolete, and offer zero value. Google could (and will) do what you do.

Bad SEOs create bad AI content, and good SEOs create good AI content. Bad SEOs don’t see the clear mines that AI leaves in your content, that you need to sweep before that article can go live.

Good SEOs see those mines, and also design prompts and structures that heavily reduce the number of mines that generate anyway.

So, I don’t think AI content writers threaten good SEOs too much. If you’re battling over prompts beyond technical SEO skill, then that’s just a modern SEO battle. And if you can win on this, then you’d probably outrank me anyway in the SERP if AI didn’t exist. It’s fair game.

I’m not a big AI guy – we haven’t stopped using any writers on the team yet as a result of AI. But it does a lot of things better than humans, and at a fraction of the price, and so it is essential to build it into your processes, and look for ways to automate what you required humans for before, so you can be as effective as possible.
​

Hope this helps!

​
​

Jamie I.F. 🔥

​
​

​

​