7 Ways to Use AI to Help You Find New SEO Keywords

Whenever any new tool or technology hits the scene, people are eager to make use of it for their jobs, whatever those jobs are. For we marketers, generative AI has been a blessing and a curse. There are certain tasks that AI makes much, much easier. Other tasks are made more difficult with AI in the mix, and some of the ways it has been abused have cost people jobs, livelihoods, and previously thriving websites.

It’s critical, then, that you exercise caution when you’re considering using AI tools. Using them the right way can save you time and possibly money. Using them improperly can result in a site being completely removed from the search index.

Using AI to help you generate keywords can be a potentially useful option, especially when it comes to PPC and more organic SEO. However, there are a few critical flaws with current generative AI that it’s important to be aware of before you start using it. There’s also an important distinction to make between blog topics and website keywords, that can be important. So, let’s dig in!

Keywords Versus Topics

Before I even touch on AI, I want to make a distinction and lay the groundwork for the rest of this discussion.

When talking about marketing, content production, ads, and similar topics, people talk about, well, topics. Topics and keywords, specifically. Often, they use these two terms interchangeably. I’m guilty of this myself in situations where the distinction isn’t as meaningful.

A topic is the subject of a piece of content. The topic of the post you’re reading right now is “using AI to find keywords.” That topic is rather broad and general in that it allows for a lot of discussion with different sub-topics and different aspects of the overall topic. AI can be great for developing topics, and I’ve covered that in detail here.

A keyword is much more specific. Something like “Keyword AI” or “New SEO Keywords” would be keywords. Well, key phrases, but the difference between a keyword and a keyphrase is meaningless in marketing today.

It’s important to remember this when we’re talking about specific keywords and not more general topics. Generative AI can be good for coming up with topic ideas or taking ideas you already have and spinning them in a way that helps you think laterally about them and synthesize new ideas. It may be less valuable for specific keywords, but that’s something I’m going to get into throughout this post.

So, how can you use AI to help with keyword research? Here are seven methods and options you can try out.

#1: Use AI to Transform a Keyword into a Cluster

Keyword clusters are groups of related keywords that can be used for a similar or the same purpose. With PPC advertising, you can use a clustered group of keywords as keyword variations to target with broad or phrase matching. With organic SEO, you can use the keywords in a cluster throughout a single post to ensure that you cover as many variations of the topic as possible and that your content can rank for hundreds of different keywords.

Generative AI is broadly capable of generating keyword clusters based on a single keyword or a short list of keywords. All you really need to do is go to one of the AI platforms and ask it to generate keyword clusters out of your input.

For example, I asked ChatGPT to generate keyword clusters based on “AI keyword research” as the base keyword. It gave me ten lists of five keyword phrases each. Feeding one sentence into an AI and getting 50 keywords back seems pretty good!

The trouble starts to come when you investigate the keywords it generates. Some of them are fine; AI-powered SEO tools, machine learning for SEO, AI keyword planner, and so forth, these work as keywords.

Some of the results it gives me, though, are clearly topics and not keywords. Things like

• AI keyword research vs. manual keyword research
• How AI improves content relevance
• How AI analyzes keyword trends

These are topics. You can use them as search queries, but you wouldn’t target them as the keyword in PPC advertising. This is a flaw that means you can’t just copy and paste AI results and use them as keyword lists, you have to filter through them manually.

#2: Strip Mine the Keywords from a Competitor’s Content

Previously, the AI platforms didn’t have much of any way to reach the internet at large. You could conceivably copy and paste a piece of content from the web and add it to a prompt, but engines like ChatGPT have a character limit on their prompts. ChatGPT’s limit is 4,096 characters, which is a little less than the post you’re reading up to this point. Since many pieces of content on the web are much longer than that, you wouldn’t be able to do whole pieces of content at once.

Modern ChatGPT, and some of the other AI platforms, do have the ability to look at the web. You can take a piece of content from your competitors, and use the URL to ask the AI to give you the top keywords from the content.

While this is handy, it’s not necessarily very deep. In the limited testing I’ve done of this method, while it has a pretty good hit rate of generating keywords rather than topics, all it tends to do is pull the keywords from the subheadings used in the post. That’s better than nothing, but you don’t really need an AI to do it for you; any web scraper can do the same thing.

#3: Evaluate Search Intent with AI

I’ve mentioned extensively throughout both of my marketing blogs that one of the most important elements of modern SEO is matching content with keywords with user intent. There needs to be a solid through-line between these; otherwise, the jump from query to keyword to content won’t happen.

Think about it; if you search for pizza recipes and the search engine gives you a bunch of pizza restaurants, you aren’t going to be happy, right? You want to make a pizza on your own, not buy one. The same is true in reverse; if you search for pizza nearby, but all you get are recipes, you aren’t going to be satisfied.

The AI tools you use to find keywords can also potentially give you some insight into the intent behind the keywords. Why might a user be searching for those keywords?

The issue I’ve found here is that the AI will give you a bit of meaning behind a keyword, but it doesn’t neatly classify them by intent. It won’t tell you when something is informational versus transactional; it will just tell you that a user is seeking a particular kind of information, which you need to interpret as a given kind of intent.

#4: Generic Basic Keyword Lists

One of the simplest options you can use generative AI for is to generate lists of keywords for you to use. A big part of keyword research is usually using tools to take a basic topic and distill it down to keywords or simply brainstorming dozens of keyword ideas, which you can then plug into other tools to evaluate. AI can offload a lot of this brainstorming for you.

This is one of those techniques that have the potential to be very useful and time-saving, but it can also be a huge trap. The AI doesn’t have any idea what a keyword is, not really. But, it does know how words are related to other words because that’s the whole way the AI functions. The problem you’ll run into is that the keyword lists the AI generates for you may be full of good keywords but equally full of bad or even nonsensical keywords.

One of the biggest issues with using generative AI for generating content is that AI works by giving the most likely strings of words to match a given input. The more likely it is, the more likely it is that it’s essentially a reflection of what someone else has already created, so it’s not really unique or valuable. For raw content, that’s terrible. For keywords, you’re actually looking for the keywords most commonly used so you can evaluate their competition and target them. In that sense, you’re almost using the AI’s own restrictions for your benefit.

#5: Geographic Keyword Variation

Another mechanical use of AI for keyword research is taking a more centralized, generic keyword and using the AI to create regionalized variations of that keyword.

Truthfully, this isn’t something you need the AI to do for you. But, it’s one of those tasks that can be tedious and time-consuming to do manually, which is where AI can offer the most potential benefit. The trick is that you might not always need a bunch of different geographic variations, so this is the kind of technique only certain kinds of businesses will need.

#6: Analysis of Your Data

Here’s one where I confess to being in the dark on, as far as personal experience goes. Many people mention the ability to use AI like ChatGPT to take data you feed into it and analyze it in a number of different ways. Some people claim it’s incredibly accurate; others are skeptical. I’m on the skeptical end for two reasons.

The first is that the large language models are not data analysis tools, nor were they designed to have those functions. A lot of data analysis is just math, and there are plenty of existing tools that can take your data set and make it valuable in different ways. Large language models are used to generate realistic-looking results, but without verification, how do you know those are accurate? There’s a fundamental lack of trust in something that can generate pretty lies with no indication otherwise.

The second skepticism I have is that I’ve seen a lot of reports that the LLMs take the prompts you use and further use those prompts and their responses to train the next generation of AI. That alone might not be a big deal – even platforms like Ahref and Semrush use customer data, properly anonymized, to perform research studies. The trouble is that there are a lot of instances of people being able to extract training data, including confidential information, from the LLMs. I don’t trust companies like OpenAI to have any regard for customer data, so I hesitate to give it any.

#7: Use an AI Keyword Research Tool

There are really two ways to approach using AI for keyword research. The first is to take an LLM tool like ChatGPT and run through various prompts with inputs and data, aiming for the goals I’ve listed above.

The other is to simply use one of the many modern marketing tools that have decided to add AI to their features list. Even large-scale tools like Ahrefs and Semrush use AI somewhere in their back end, though they don’t necessarily specify how or why. Truthfully, I think a lot of it is just a rebrand of the same MLM algorithms they were using before; the current generation of LLM AI isn’t really suited to the more mathematical and reliable tasks.

The other benefit is that these tools address one of the biggest problems AI has for keyword research: a lack of reliability and a lack of data sourcing.

When you do keyword research, you need more than just a list of keywords. You need data to go along with it. You want traffic estimates, you want competition estimates, you want difficulty analysis. An LLM can give you this, but from where? They don’t have data sources. AI tools get around this, but so do standard keyword research tools.

When you need to find actual data from a reliable source, you don’t want an LLM that can hallucinate in a way you’d never know. You need a trustworthy app, whether it’s something like Topicfinder, Ahrefs, Semrush, or even just the Google Keyword Planner.

Leave a Comment