How To Use AI For SEO (7 Proven Examples With Custom Prompts)

You know SEO is crucial for bringing qualified traffic to your site, yet you struggle to find time to tweak on-page elements, build external links, and create well-optimized content.

What if you could use AI tools to optimize and scale your SEO efforts 10x, freeing up more time for your high-impact business activities?

In this guide, I’ll show you my insider secrets for how to use AI for SEO (the right way that won’t harm your site). I’ll also walk you through 7 specific AI SEO use cases and share examples of my custom prompts.

I’m a content strategist and SEO coach with over 10 years of experience in online content marketing. I’ve grown two profitable blogs on my own with just AI and SEO (no paid ads and no team of writers).

I’ve also contributed expert opinion to industry-leading websites such as Moz and HubSpot, and been featured in Business Insider, the Financial Times, and TEDx, talking about AI, SEO, and online disinformation.

What Does it Mean to Use AI in SEO?

Over the last 12 months, I’ve noticed a lot of misunderstanding about the role of AI in SEO, especially when it comes to content creation. Upon reflection, this boils down to two main issues:

1. People using AI incorrectly – e.g. bad prompting, using AI where a human should be used
2. People using AI unethically – e.g. stripping a competitor’s sitemap then using AI to autogenerate articles for all URLs (this “one-shot” approach to AI in SEO is hugely problematic in my opinion)

In this article I want to teach you a different approach to using AI in SEO – one that treats the tool as a smart assistant rather than a replacement for human writers.

This is an approach that augments human expertise with AI tools, rather than replaces it. And it’s one that takes the time and has the patience to prompt carefully and intelligently, rather than leveraging shortcuts that fill the internet with yet more junk content.

Here’s an example with a basic prompt (I’m using Claude 2.1, the previous version. It delivers more natural output than Opus).

As you can see, the ChatGPT version is full of over-the-top metaphors, flowery language, and obvious “AI-isms”. These are a dead giveaway that something has been written by AI (especially now that internet users are getting used to seeing bad AI content).

On the other hand, Claude’s basic output is far more human. I still wouldn’t use this text without human editing (and I would give Claude a more sophisticated prompt to start with). But it provides a solid starting point for further content creation efforts.

7 AI SEO Use Cases (with Example Prompts)

#1. Generating Endless Content Ideas

Publishing fresh content is the backbone of an SEO strategy. To have the best chance of attracting organic traffic you need to produce a regular stream of high quality, well optimized content.

But many entrepreneurs ask me “how do I know what to write about?”

For starters, you should always talk to your audience and map those pain points onto your keyword research process. That’s a really good start.

AI tools can also help you generate endless lists of content ideas. They’re the perfect sidekick when you’re seeking an unusual or surprising angle on a popular topic. I normally do my keyword research first, then give the target keyword to my AI tool and ask it to generate a list of unexpected angles on the topic.

Don’t forget to also look at what your competitors are doing in the top 10 search results, and use that as a guide to craft your article. Often, ranking for a certain keyword calls for specific ‘ingredients’ (i.e. sections), although you should still have sufficient room for creativity and your unique viewpoint.

#2. Creating Intelligent Outlines

You should always outline your blog post before you start writing. In the recent past, a lot of SEO advice suggested looking at the top 10 search results and creating an outline based on key sections your competitors are using in their own articles.

But after Google’s September 2023 helpful content update, that’s no longer the main rule of thumb you should use. In many cases, that approach tends to produce regurgitated articles that don’t have enough differentiation from competitors. Google doesn’t want that.

Nevertheless, you should still take some time to look at what your competitors are including. For example, perhaps your target keyword calls for an introductory section explaining what the topic is all about.

If every top 10 competitor has this section in their article, then it would be a bad idea to skip it in your own. But use this as a guide, rather than emulating it too closely.

When creating outlines, AI tools are great for helping you get past writers block and come up with imaginative ideas to include in your article. I recommend using a combination of competitor research, AI support, and your own subject matter expertise to create winning outlines.

#3. Crafting Clickworthy Titles and Meta Descriptions

Writing titles and meta descriptions for your blog posts can be a tedious task. But they’re extremely important for getting people to click through to your articles. Fortunately, this is a great area for AI tools to step in and take the strain. Summarizing content is one of their major strengths.

For titles, I simply give my main target keyword to the AI tool and ask it to generate a list of title ideas. If I don’t like what it generates first time, then I just explain exactly what I don’t like and ask the tool to try again with different parameters.

For meta descriptions, I once again give the tool the target keyword and ask it to generate 10 ideas for compelling meta descriptions of no longer than 160 characters. Sometimes it gets the number of characters wrong, but you just need a few manual edits to make things perfect.

#4. Drafting Introduction Hooks

It’s helpful to start your blog post with a compelling hook, to draw the reader in and make them read on. At the same time, you don’t want to overwhelm them with a long waffly introduction.

To get a good introduction hook, I give the AI tool the main keyword and ask it to draft a compelling and unexpected hook using the problem-agitate-solution copywriting framework.

Sometimes (if it’s not immediately obvious) I also ask the tool to give me ideas for different angles on the problem that my target audience might be facing. I then use this information for crafting better hooks using the PAS framework.

#5. Turning Keyword Lists into Topic Clusters

It’s no longer enough to target single keywords in isolation. For best SEO results with your blog, you’ll need to build topical clusters with your content. But often, the mechanics of this process aren’t very clear. Luckily, I’ve found an easy solution that gives good results.

In this case, I export a CSV (.csv) file of related keywords directly from my favorite keyword research tool (Semrush). Next, I upload that file into Claude and ask it to organize them into semantically related topic clusters, then present that information in table format.

Then, all I have to do is create and publish the individual blog posts, making sure they contain a comprehensive internal linking structure to tie the cluster together.

#6. Evaluating Topical Authority

Topical authority refers to the level of expertise and credibility your website has on a specific topic.

You can achieve topical authority by consistently publishing high-quality content covering your topic from a wide range of angles. Over time, your website becomes known in your industry as a go-to source for valuable information.

What’s more, Google likes topical authority too. The more informative and high-quality articles you publish on your blog, the greater the chances of it reaching topical authority and being rewarded with better Google rankings.

But it’s easy to stray off track and end up publishing content about topics that are only tangentially related to your area of topical expertise. Perhaps you’re chasing high traffic keywords. Perhaps you got interested in an adjacent topic and tried to tie it back to your core topic, but you’re not sure how successful that was.

I experienced this exact issue with one of my websites. I went after some high-traffic keywords (and ranked on page 1). But I wasn’t sure how well related some of the articles were to the main topic (in which I’d already worked hard to build authority). So I created a custom AI prompt to help me figure out this issue.

This prompt analyzes the entire structure of your blog and evaluates the level of topical authority. It will flag any posts that may be diluting your site’s topical authority and suggest ways to revamp them to tighten up the focus.

For input, you’ll need to copy and paste your website’s full sitemap into a file, then upload the file into a new chat window in Claude. Then enter the following prompt.

#7. Imitating Your Personal Writing Style

Many content creators worry that using AI tools will dilute their personal writing style and they’ll end up with content that sounds overly robotic or fluffy. That’s certainly a valid concern, especially with so many people using vanilla ChatGPT with basic prompting. Here’s an example with a very basic prompt.

I’ve developed several tactics to overcome this problem. One way to do it is to first outline your article, then use the AI tool to fill in each section one by one. Then, you go over each section yourself and edit it to match your personal style.

But this approach can be cumbersome and time-consuming, Plus it’s not always clear how to edit sufficiently to make the text sound like your own writing.

I like to teach Claude how to imitate my personal writing style. Here’s how I do it. First, I identify 4 or 5 of my best articles that I’ve written completely manually. I copy paste each one of those articles into a plain text file (.txt). Each article gets its own file.

Then, I upload all the text files into a new chat box in Claude. Although Claude doesn’t (yet) have access to the Internet, its document upload feature is very powerful.

You can upload all kinds of documents (including complex ones like reports or white papers) and ask it to perform prompts that synthesize and interact with the information in the documents.

Once the documents are uploaded, I first prompt Claude to read them and generate a description of the style, voice, tone, and structure of the writing. It normally produces a set of bullet points for each writing feature, which combined provide a pretty accurate summary of my writing style. Now Claude has learned how to write like me.

Next I ask Claude to generate a new section of text using exactly the same style, voice, tone, and structure as it just summarized from reading the documents. This allows me to evaluate how well it has imitated my personal style. And it usually does an excellent job.

Now I can go forward and use the output of this chat to generate sections of text in a close approximation of my own writing style. As always, if you don’t like the output, just write another prompt in natural language to tell Claude what you don’t like and suggest what it should change.

Best Practices When Using AI Tools

• Treat Your AI Tool Like a Human – Talk to it as you would a human assistant. You don’t need to use fancy or technical jargon to make your AI took work for you.
• Don’t Just Prompt and Go – You’ll probably need to interact with the AI via several prompts before you get to the output you’re looking for. Make sure you have the patience to do this.
• Provide Sufficient Context – Give your AI tool extra details, such as what you’re trying to achieve, what you’ve already done, and any other details to help it understand the background of the problem.
• Provide the Right Input Materials – Sometimes your AI tool needs specific inputs to give you the best results. For example, if you want it to imitate your writing style, you’ll have to give it a few samples of your own writing.
• Find a Proven Structure to Follow – I recommend finding expert structures to plug into your tool to help it get you the desired results. For example, if you want to create a great YouTube video script, you should look for an existing framework for YouTube videos that go viral. Use this framework as your input and ask the tool to follow it.
• If Output is Wrong, Say So – If your tool gets something wrong, let it know. And tell it what you want to change.
• Interrogate Your AI Tool – Don’t take everything your AI tool says as fact. Be prepared to query it at any point if you’re not sure about some of its output.

Prompting Claude AI (Quick Video Tutorial)

Here’s a quick video walk-through showing me demonstrating a prompt in Claude.

Actionable Next Steps

Learning how to use AI for SEO opens up endless possibilities. The 7 use cases I’ve outlined in this guide will give you an insider playbook to streamline some of your SEO workflows, beat your competitors, and win more qualified organic traffic.

The key next steps are to:

1. Identify your top SEO pain points and assess which AI tools best address them. I recommend using Claude and starting with content creation and initial keyword clustering.
2. Carefully craft your prompts using my examples as templates. Remember – provide enough context and guardrails so AI augments rather than replaces your domain expertise.
3. Design an SEO campaign leveraging AI around a new product launch, site upgrade, or priority business objective. Establish processes to keep refreshing your site’s content and keywords.
4. Track analytics dashboards and rankings closely to compare your site’s performance pre and post-AI integration. Look for tangible traffic, leads and sales wins.
5. Continue optimizing your prompts and workflows. Set youself goals to use AI for one additional SEO task every quarter to further accelerate your growth and learning.

SEO is a perpetual process, but AI makes constant innovation and growth more sustainable. You now have the insider techniques to work smarter – and take your business to new heights!