Welcome to this new era of playing with AI Tools and using AI-generated content!

It’s all the rage right now which is exactly why guides are the new “must”.

As a web design and digital marketing company, we’ve had our hands and eyes on these tools even before they blew up in the news. Development companies like ours had already experimented with AI programs to create certain plugins, and many coders have been using AI platforms to debug issues for years.

But it’s now a whole new ballgame to use AI to create content. That’s what we want to dive into in more detail here.

We want to talk about the precautions in relying on AI to create content and what you need to do if you want to use it.

So if you’re already using AI Tools or considering it, you’ll want to take note of these lessons that we learned from our experience with it.

AI-Generated Content DOs and DON’Ts

• DO use it to gather new ideas on topics, and to expand on content you already have.

• DON’T copy and publish content without doing fact-checking and heavy editing.

• DO use it to create social media post copy when you’re out of original ideas.

• DON’T use its hashtag suggestions without double-checking them all on the platforms first.

• DO use it to create email marketing content; given that you provide specifics in the prompt.

• DON’T ignore any flags and suggestions prompted by your email marketing platform on how to format content.

• DO ask for several responses on the same prompt to get samples of content blurbs to work with.

• DON’T rely on it to know what needs to be done for SEO, even when you ask it to use high search volume keywords.

It’s true that many are on either side of the fence when it comes to whether these items are DOs or DON’Ts. No one is necessarily considered wrong, but it appears to be a matter of perspective that depends on the use and experiences. For an additional read that explains why many SEO Specialists in particular are cautious of AI-generated content, check out this blog post as well.

What We’ve Learned From Using AI Tools for Content

First and foremost, you need to train it before you can start getting usable content. It’s seemingly straightforward enough to get what looks like usable content in just two clicks. But if you’re trying to publish a web page or blog post, you may very well end up with worse results than what you had when you started.

Why?

We’d rather share some facts from our experience here…

When we started experimenting by using ChatGPT to aid our content efforts, we asked it to tell us:

1) What keywords needed to be used.

2) Blog post topics to write about.

It produced a list of 25 seemingly relevant keywords. But when we did our research on this list using SEMRUSH, we didn’t find a reason to use any of those phrases. In fact, there weren’t even any related suggestions that we could use.

Unfortunately, we also had similar findings when we used the blog topic list. Upon doing the research on both questions [asked in Google] and in SEMRUSH’s Topic Research tool, there weren’t any relevant matches.

The best it did was give us a starting point for research on both keywords and topics. So we can credit it with that much… but still note the precaution against the whole “copy and paste” method.

Some of our clients who were short on time have used AI Tools to create their blog posts. But without taking the steps that we’re going to talk about further down, they presented us with content that didn’t even meet the minimum requirements for SEO. This resulted in a need for us to rewrite it and send it back for review before we could begin drafting it on their website.

The bottom line was that if we relied on what ChatGPT told us to do, we’d be producing content that very likely wouldn’t be looked at. And since the purpose of our services is in delivering content that gets found online, relying on that would’ve caused negative results.

The Warning on AI-Generated Content and SEO

As an SEO Specialist, this has admittedly been a bit of a love/hate relationship. As mentioned, the keywords provided by AI Tools so far haven’t had any encouraging data to support their use.

On the other hand, we’ve learned to do our research before coming to the AI tool, and then telling it what it needs to write about and how to write it.

While it seemed like this process was a surefire way to create great content, we learned from this too…

After taking that content and running it through the other tools we use for optimization, we found it still set off a handful of flags. It wasn’t as bad as generating content without the research and requirements inputted, but it still needed further editing to get the green “SEO-friendly” light.

The truth is that while we hoped AI-generated content would save us time, it took a lot of time to check, edit, and optimize the content. In the end, it took almost as much time as it normally takes us to write a blog post from scratch without AI.

The biggest SEO flags we’ve seen (that you should look out for) are:

• • Keywords were underused when not specified and often overused when specified.
    Since the # of min/max keywords is based on the total word count of any given page, the only way to know is to use an SEO plugin.

• • Keywords were not at all present in headings even when prompted to do so.
    Headings are a must for Accessibility purposes, but also important as a means to support the use of the keywords that apply to the page.

• • Word counts were always lower than expected (which varies based on the topic/headline) unless it was told to meet “X word count”.
    Note that all pages aren’t considered “SEO-friendly” unless they meet a 200-word count minimum, but 300+ is better. Blog post SEO works a bit differently and is measured by the topic; so the only minimum we can speak to as a general rule is 500+.

• • Certain phrases appeared in every single blog post and became known as dead giveaways of AI-generated content. Examples are:
    “In this blog post..” in the first paragraph,
    “In conclusion” or “In summary” at the beginning of the last section/paragraph,
    and repeated use of “Additionally”, “Furthermore”, and other words like those to start sentences.
    These are examples of wording that should be rewritten because they aren’t considered best practices from a Readability standpoint.

• • Content from previous posts/prompts would be duplicated unless you specify not to with every prompt.
    Duplicated content is another necessary thing to avoid when it comes to SEO, so be sure to always say “Do not repeat any sentences previously used”.

So then why use AI Tools for Content?

Even with all that being said, AI can very well be used as an aid. But since it is a tool, it needs to be trained on how to speak your language before you get anything to use.

How you do this is educate it on your brand, target market, and any and all lingo (ie. slogans, examples of brand voice) first. While it seems redundant, you should also ask it to ask you questions before giving any prompts.

We literally start by inserting whole copies of bios and service listings, along with Vision and Mission statements, and details on the target audience, then ask it what else it needs to know to create content for that company.

Every time we do this, it asks us to explain all the other “ins and outs” of the business to get started. The plus in spending the time doing this is that it adapted to the brand and spoke more to the target audience now that it has noted it.

This also helped to steer away from the very factual and robotic/monotone responses it otherwise gives. Because the other unfortunate downside with so many people now using AI-generated content – but perhaps not editing it – is that at some point, something will set off a plagiarism flag. In fact, we’ve seen blog posts on different topics, each from very different companies, that contained the same phrases because it was generated by AI without the training step.

So it’s only a “win-win” to put in the time to ensure the focus and structuring of the content writing.

Second to that is the crucial warning to heavily check and edit the content it produces. While it has been helpful for us to produce content for clients in industries we know nothing about, the following has happened:

• It wrote a blog post stating something was illegal – which isn’t at all illegal. This was despite specifics and location data supplied in the prompt.

• It wrote “how to” posts on manufacturing processes which turned out to be inaccurate – the client had to remove whole sections of blog drafts because steps didn’t apply.

• It wrote the exact same explanatory content for several different topics and clients – which only added more questions that made us scrap both articles. This was also despite the fact of training these channels to be specific to these clients.

• When we used a plagiarism checker for a blog post, it showed us that it completely reworded existing content. We probably would’ve overlooked this otherwise, especially since the flag came from the fact that it “forgot” to reword two sentences – they were exact word-for-word copies from the other website.

These are real examples of why you should also complete genuine research, or, know enough about the topic matter in order to verify the content.

If relying on getting content for SEO purposes, be sure to do your research first and specify those keywords and parameters in the prompt as well.

But once you’re in the habit of doing all of this, then you can use the content it provides. Even still, read through it a couple of times and do some tweaking to make it sound like it’s you. If you don’t, you risk having your loyal client base & fans possibly seeing through it much like we can now.

This blog post was written (without the use of AI) by Northnet Media’s SEO Specialist and is based on actual experiences. At Nothnet Media, we strive to be your all-in-one online strategy with a fully optimized website, plus campaign landing pages, ads, and blog posts. Contact us today to discuss your business goals and we’ll tell you how we can help you achieve it!