Using AI for Keyword Research

Digital Marketing - Study Notes:

Keyword research isn’t just about using conventional tools and manual brainstorming. AI can help by delving deep into the search intent of keywords, brainstorming new and emerging topics, and identifying content gaps on your site.

Analyzing SERPs

When it comes to SEO, one of the biggest mistakes even experienced SEOs make is aligning content with the wrong search intent. This often stems from misguided keyword research.

For example, if you target the phrase “best running shoes for flat feet” and review Google results, you’ll find it has a commercial search intent, falling between informational and transactional intent. Targeting this phrase on a purely transactional e-commerce page would make it nearly impossible to rank for.

However, the four main categories of intent - navigational, informational, commercial, and transactional - are broad, and the search engine results pages (SERPs) are often more nuanced than that.

One solution to that is Ahrefs Identify Intents feature. It’s part of their paid plan, so you start with a keyword in their Keywords Explorer, such as “best running shoes for flat feet.” Then, scroll down to the “SERP overview” section and press the “Identify Intents” button.

Using AI, Ahrefs will analyze Google’s SERP for the given term. It will go beyond the broad search intent and provide insights into the specific types of content that are ranking.

For instance, on the right of the screenshot, Ahrefs indicates that 55% of the search intent for “best running shoes for flat feet” is related to product recommendations, suggesting commercial intent.

However, 31% are seeking expert advice, and 9% through video reviews.

What can be learned from this? If the percentage seeking expert advice were significantly higher and your site lacked that type of content, then targeting this keyword might not be beneficial. However, if your site already features a product review article that struggles to rank, it may be more effective to try a video review and embed it in your article. This could potentially double your exposure and attract traffic from both Google and YouTube.

Let’s look at another example, for the term “how to start a blog”.

We can see that 46% are looking for a step-by-step guide, which is not surprising given the “how-to” nature. However, we can also see that 27% are looking for beginner-level questions, while 16% are seeking community discussions and advice and 11% making money through blogging.

So, even though the search intent is broadly informational, the specific intent is more nuanced. Armed with this information, bloggers writing about blogging may decide to create content tailored towards the larger market and produce a step-by-step guide geared towards beginners, or they may choose to aim it more niche, at an intermediate level, slanting it more towards a monetization format. Both approaches have a good chance of ranking for the term “how to start a blog”.

Brainstorming trending topics

When brainstorming trending topics, Google Gemini’s ability to access up-to-date information makes it our preferred AI language model.

For instance, assume you’re a blogger at a sustainable materials company seeking up-and-coming and slightly unconventional content inspiration.

So your prompt could be:

List five separate trending topics for sustainable materials currently gaining traction and are a bit unconventional. In the brevity of a meta description.

Gemini comes back with five interesting results:

These topics could serve as interesting angles for individual or combined blog posts. As they are trending, they would likely perform well on social media or through your email newsletter. Using Google Gemini, alongside Google Trends can help with novel and current content ideas.

Clustering topics

Google Search Console provides a wealth of data, like telling you which keywords people are using to find your website – but it lacks the ability to organize these keywords effectively into topic clusters. This is where using AI can be a game-changer. We’ll use ChatGPT, but you can also use Gemini, Claude, or another AI language model.

The first step is to export your keyword data from Google Search Console. So, visit the “Search Results” section under “Performance”, then click Export in the top right corner and select your preferred format, such as “Download Excel”. This file contains all the keywords, or “queries,” that your website has ranked for on Google in the last three months.

Once you have your list of keywords, the next stage involves feeding them into ChatGPT. You’ll need ChatGPT 4o or above as this allows you to upload you Excel document by clicking the attachment icon.

You can then use the following prompt:

Review the queries from Search Console. Group them into topic clusters where they share the same or very similar search intent and can be served by a single page.

The AI can analyze these keywords and assign them to different topic clusters based on their semantic similarities and search intent. This method not only simplifies the organization of your keywords but also provides a clearer insight into the themes and subjects that are driving traffic to your site.

For example, let’s say you run a gardening blog and you’ve extracted a list of 1000 queries from your site.

ChatGPT might identify topic clusters such as:

• Indoor Gardening (26 queries)
• Gardening Tools (19 queries)
• Bee friendly plants (14 queries)
• Seasonal Plant Care (9 queries)
• Sustainable Gardening Practices (14 queries)

Now, there’s a fair chance the AI only provides a sample of topics so you can provide a follow up prompt.

Please provide all topic clusters as a downloadable CSV.

By asking for a CSV file, ChatGPT tends to provide more data than it’s regular interface. If you want to see the queries per cluster, you can try, but depending on your data size, this might stretch ChatGPT’s current processing capabilities. If that’s the case, try a smaller sample at a time.

For example, here’s a prompt you could use for that.

For the topics below, please list the queries that fall into each group.

Indoor Gardening
Gardening Tools
Bee friendly plants
Seasonal Plant Care
Sustainable Gardening Practices

Identifying these clusters allows you to compare how well these topics are represented on your site. For instance, if the “Sustainable Gardening Practices” cluster has high search volumes but limited content on your site, it highlights a content gap, which presents a prime opportunity for new content development.

Additionally, here’s a prompt for ChatGPT to discover new content gaps.

After reviewing the full list of topics, identify content gaps for related queries where the website currently ranks for very few or no queries.

Furthermore, understanding topic clusters aids in crafting content that not only fills these gaps but can also help with interlinking strategically. This enhances user engagement by providing a more cohesive learning path and improves your site’s authority on specific subjects, thereby boosting SEO performance.