Heyyyy,

Now the newsletter is weekly – so expect more of these every Thursday.

This one’s my philosophy on AI content, where it works, where it doesn’t work, and why I think most people have a too-binary opinion on it.

The truth is it’s nuanced.

–

AD:

From 0 to 189,000 Monthly Organic Website Visitors

SEO Content Writing Subscription Service

Adam Crookes (Freshly Squeezed SEO) recently took a customer’s domain soaring into the stratosphere with monthly blog content.

Freshly Squeezed SEO will help you outrank AI content by delivering helpful, publish-ready blog articles that meet Google's E-E-A-T guidelines.

When you subscribe to Freshly Squeezed SEO, you are paying for the "competitive edge" that will catapult your domain higher up the SERPs.

​WATCH VIDEO: Adam’s SEO Content Writing Subscription Service​

–

My Opinion on AI Content

I think if you follow me you’ll assume I’m anti-AI.

I’m not overall – I already use it. But I think most people use it wrong, and they just haven’t been punished for it yet.

First things first, AI is not going away. It’s the final boss. And I think we’re all fucked in the future (but nihilism aside, I’ll keep this email to the point).

It’s not some Web3-esque flash in the pan. AI already has utility and already saves me time and makes me more productive.

For example:

I can paste any article into GPT-4, ask it to create a summary of this article, and it’ll usually do it perfectly (still check it, obviously). This is very helpful for:

• VS articles: it’ll summarise the differences between two products across each section, giving a quick summary and helping readers decide which to buy.
• Guides: it’ll give you the key takeaways, and even summarise each step concisely.

I even have a confession to make.

Recently, at Lasso we decided to update a lot of our “best affiliate programs for X niche” articles. They’re big lists of all the main affiliate programs e.g. in golf, fashion, jewelry, finance, etc.

We already have collected the data from 10,000+ affiliate programs and we display this programmatically, but we weren’t ranking because a lot of the data and affiliate programs are not that semantically related to the main keywords we were targeting.

We needed a bit of old-fashioned H2 and keyword/entity stuffing!!!

So, now all these pages contain a templated list of FAQs, and H2 headings, that are entirely AI-generated within this programmatic template.

The result? We’re mooning:

TLDR: AI, when used properly, works. We were fine using AI here because we had useful info (we collected all the aff program data), and AI just augmented the presentation of this data collection.

*** Also, if you’re a big programmatic fan, I’ve got an announcement very soon that you’re going to love. It’ll change how you monetize your programmatic pages and help you make serious $$$ 🤑

–

However, I still hate the philosophy a lot of people who use AI content embody: wanting to do the least, and get the most.

AI draws out all the chancers who want to get rich quick with no work.

And that won’t happen. I hope you enjoy the next Ponzi scheme that gets offered to you that promises to grant you instant wealth.

AI is good, for some things. It’s not AGI, and it will not print you money unless you have an already profitable skillset.

AI is good for objective, manual, labour-intensive things. Creating tables, summarising data, or adding entities that you specify into your paragraphs.

It’s also great for finding what I call “connective tissue”.

If you paste in a paragraph you’re just not fully happy with, and feel like it’s not informative enough, with the prompt:

“Is this correct: {content}”

It will confirm it’s correct (if it says it’s wrong, you have a bigger problem)...

…and it’ll also say either where things can be made clearer, or where you do not give the full info on a topic, which can quickly inspire ideas for small parts you can add to make this section better and more comprehensive..

However, it is good at these things because they are very siloed, directed, and shallow tasks.

If you give a more wide-ranging prompt, you’ll get a messier response. There’s too much noise and subjectivity to get you what you want.

It Augments, Not Replaces (So Far)

Because AI doesn’t perform well with open-ended prompts, it cannot scale you to the level where you make enormous money without you also being talented enough to manage it at scale.

I am not worried about AI content – or at least, pure AI content – beating me in the SERPs.

Because bad SEOs create bad AI content, in exactly the same way that bad SEOs create bad human content.

For exactly the same reason: they haven’t learned and gained the intuition for what does and doesn’t work for ranking on Google.

And with AI content, it’s a minefield. It often goes off-topic, lies, makes up data, and if you don’t manage it well as a content manager, you will not get results.

But good SEOs can harness AI to create good content at scales they could not previously.

And so if you beat me in the SERPs with partially AI content, fair play to you. You were probably talented enough to beat me anyway in an AI-less world. AI doesn’t replace talent, and doesn’t replace people – it augments talent, and amplifies the bad if you’re untalented.

This is why I think the winners from the AI content wave are not going to be things like Jasper, or even the GPT APIs.

It’ll be tools like Reword, that harness AI content but use them with other forms of data to give you a bird’s eye view of what human things only you can do, to get the best results. And then automates the rest.

For example, if you’ve ever played poker online, you may have heard of a “HUD”.

HUD means head’s up display, and basically, they’re software that overlays all the players on your online poker table with a bunch of stats. The percentage of hands they voluntarily play, how often they 3-bet, how often they fold in the small blind, etc.

With this data, you can tweak your strategy to win the most against the tendencies your opponents have.

Tools like Reword (and, to an extent, SurferSEO) are basically the SEO/content version of this. I’ve been playing with this a lot, and yes it is like a standard AI writer like Jasper et al, but it's also:

1. Based on your existing body of work (it learns how you write from your content, and writes like you do, and your specific knowledge)
2. Adds a data-driven “HUD” over the top. With one click, you get access to all the statistics available about the topic you’re writing about to cite, can get recommendations for internal links or external links to add, any FAQs you should add – so you can take the human decisions that make content good or bad. But the info is fed to you in record time and built off of AI.

​

This adds to productivity, and augments talent as you can decide which things you want, or don’t want, based on your intuition and skill of what works.

Though, I'm still barely using any AI on my main sites. I have a team of great writers who can produce way better stuff than any AI can, and it's all hand-written and for users.

Remember:

AI cannot (and should never) hold strong opinions on anything. This should be a uniquely human thing. And so, good “AI” content should be improved still with those human elements, unique hands-on experience, and emotional opinions that advise and connect with people.

And if you think you can click a button and generate a site and make money forever, what do you really think your unique selling point is? What makes you different from everyone else who has access to the same technology?

One-click can work in a few situations, but all of these involve unique talent or money investments to make it less easy to copy:

1. Aged domains – uncopyable links, which makes copyable content perform in a less copyable way
2. Extremely strong topical map – a very difficult-to-copy technical skill, can make for authority that is more difficult to replicate (low-key, you can just clone someone’s site/sitemap though if they haven’t built links)

I’m still yet to see a pure AI site that’s not dogshit.

A few sites have generated good traffic and not been smacked down (most get smacked to 0), but those sites use well-designed templates, and are basically just good programmatic SEO sites, that are augmented by AI. I don’t class this as pure AI because human design has gone into the content structure and delivery.

That leaves just 1 or 2 successful pure AI projects – and these are all getting traffic off definitions in different industries. It’s not huge RPM, huge affiliate, or huge digital product plays.

And it’s not too defensible either, because anyone can generate the same content as you – you all have access to the same GPT API.

Also, good luck finding someone willing to pay you a decent multiple for an AI site. I’d see an AI site build as a churn and burn, rather than having much likelihood of stabilising and being sellable as an asset in most cases.

So, yes, play with AI 100% – I advise you to, and to be creative. There are way more ways AI is transforming SEO and content writing beyond the few I’ve listed – and we’re using it extensively to augment our processes in-house.

I’m still learning how it can be used myself! I’m LOVING the things it’s helped us do better.

But, I think you should still see it as an augmentor, not a replacer for your unique human talents. No one wants to read AI content, they want to read human content. So don’t lose that human touch – you’ll regret it when the GSC graph starts pointing sharply downward.

Until next week 🍸

Jamie I.F.
​