Home » Keyword Research for AI Search Engines

Keyword Research for AI Search Engines

Written by Lawrence Hitches

10 min read
Posted 9 May 2025

In This Article

AI chatbots, voice assistants, and Google’s AI overviews have changed SEO in Answer Engine Optimisation (AEO). This, with the rise of conversational queries, is changing how we create content. 

Now, instead of focusing on keyword-based optimisation, we must focus on question-based optimisation. And to start this, we need to begin with learning how to conduct effective question research for AEO. 

Why Questions Matter for AEO and AI SEO 

Modern search behaviour is getting more conversational. Users are typing in keywords much less often. Instead, they’re typing in long-tail searches like “Can dogs eat toffee?”, “Why is my cat acting weird?”, etc.

AI tools like ChatGPT, Google SGE Questions, and Perplexity understand this. That’s why they’re focusing on providing clear, direct, structured answers. Therefore, if your content is formatted in this preferred AEO way, it’s more likely to get cited

Just think about it. AI tools want to provide direct answers. Instead of generating the answer themselves, which would put a lot of strain on the tool, they’d rather cite a website that already offers the direct answer. 

This is what makes question-based optimisation so important in this new era of zero-click SEO. By optimising for how users ask, not just what they search, you can boost your visibility in AI-generated results. 

How to Find the Right Questions 

To start appearing in users’ question searches, you need to start at the source: the question. There are five methods to find them. 

  1. Tools 

Nowadays, there are a number of tools used to find conversational queries. Some popular options include AnswerThePublic, AlsoAsked, and keywordTool.io.

With a lot of these tools, you enter a topic, for example, SEO, and it’ll return hundreds of related questions. The questions they mine are usually from a search engine’s autocomplete search feature and from the People Also Ask featured snippet on Google. 

These are more “dedicated” tools. However, you can also use SEO suites like SEMRush and Ahrefs, scrapers for community forums, etc. 

The SEO suites usually have features where you can search by “question” for keywords, and for scrapers, you can get them to mine popular forum discussions with question marks. 

  1. Platforms 

You can also manually find questions instead of using tools. There are plenty of Q&A websites out there nowadays. 

For example, Quora is a big one. You can go to Quora and search for a topic. In return, it’ll list hundreds of different questions that include this topic. 

Again, you can also use community forums. Review what questions are being asked and which have the highest engagement. These are the questions your target audience is interested in. 

  1. Google SERPs 

There are tools that will extract data from Google’s SERPs. But, if you don’t want to invest in a tool, you can do it manually. 

The key areas you want to extract data from are: 

  • Autocomplete: When you type a keyword in the search bar, it “guesses” what you want to search. These searches are calculated based on relevancy and frequency. 
  • People Also Ask: If you scroll down the SERPs, you’ll see a People Also Ask section. These questions have been chosen algorithmically. 
  • Related Searches: Right down the bottom of the SERPs, you’ll find related searches based on your current search. 
  1. ChatGPT Prompts 

Alongside the above, you can also use ChatGPT prompts to discover questions related to your topic, industry, or target audience. 

For example: 

By using this prompt, or something similar, you’ll get a bunch of questions related to your topic. 

  1. Customer Research

Lastly, if you already have an audience, you may naturally know the questions they’re asking. You can also perform research in the form of surveys to discover common questions they ask on a day-to-day basis. 

How to Map Questions to Your Content Strategy 

Step 1: Cluster Questions by Topic

With the list of questions you found earlier, see which can be based on topic clusters

A topic cluster has two key parts: a pillar post (broad questions) and supporting clusters (non-broad questions). 

To find pillar-type questions, find very broad questions, like “How can I brew coffee at home?”, “What is SEO?”, “What are AI tools?”, and place them into a “pillar” section. 

For supporting clusters, look for less broad questions. For example, “How can I brew coffee at home?” could have “How can I grind coffee beans at home?”, “How can I froth milk at home?”, etc. Put these types of questions in a “cluster” section. 

At the end, you should have a list of pillar questions with cluster questions. 

Step 2: Match Each Question to the Right Content Format

Now you have your cluster questions, look at the pillar and supporting clusters, and determine what type of content best serves the search intent. 

For example, “How can I brew coffee at home?” would be a great long-form post. However, it could also be a superb video. In that case, it might be worth doing both to increase the chances of appearing in Bing/ChatGPT citations. The content that the AI will extract from will depend on the intent of the question. 

However, for something a little more concise, like “Where to place my coffee machine at home”, a visual-heavy post might be better. In that case, the cluster topic would have a different content format than the pillar post. 

The most important thing is to think about how the user wants to experience this content. Does it need to be in a FAQ format, a video, an image-based post, etc., as whichever serves the intent best will be shown. 

Step 3: Align Questions with the User Journey 

The questions you develop content for should align with the user’s journey. 

For instance, someone searching for “What is coffee?” is looking for a definition of coffee. However, someone looking for “Best coffee machines for home” is looking for a list of the best coffee machines for home use. 

Similar to choosing the right content format, how you answer the question should align with where the user is in their journey. 

Step 4: Build Topical Authority Through Breadth and Depth

Topical authority is all about breadth and depth, and this is something loved by AI search engines. 

Therefore, instead of covering a topic in little depth, like “How do I brew coffee at home?”, go all in. Cover everything from “What are the best brewing machines for coffee?”, “What are the best coffee beans for home brewing?”, etc. 

AI search engines would rather extract data from one website than a hundred different options. Therefore, the more they can find on your website about a topic, the more likely they’ll cite you. 

Step 5: Create a Structured Workflow for Scaling

The very last thing you need to do is create a structured workflow to scale AEO. This is important for large websites, SEO agencies, and marketers. 

The best thing to do is to plan on an Excel sheet. Choose a topic, choose several pillar questions that cover that topic, support each pillar with a cluster, and then plan the execution. 

How you structure your answers is just as important as the questions you choose. When answering them, aim for clarity, directness, and scannability. 

Some key practices to consider are: 

  • Use proper heading structure for your headings, using h2, h3, and h4 tags. 
  • Keep answers short and direct, removing any unnecessary fluff or words. 
  • Use lists and tables when appropriate and when they answer the question better. 
  • Write “what is” and “what are” questions around 40 to 60 words long to appear in PPA. 
  • Write as you are speaking, an active, conversational voice is picked up by AI assistants and is key for voice search optimisation. 
  • Use proper E-E-A-T to increase the credibility and trust of your answers. 
  • Use a schema appropriate for the content, for example, an FAQ block with the FAQ schema. 

Boosting Q&A Visibility with Schema Markup 

Schema markup can help improve the machine-readability of your website content. And, simply, the easier it is to read by a machine, the greater the chance it’ll be cited by AI. 

Schema TypeUse CaseKey Benefits
FAQPagePages containing multiple question–answer pairsDisplays FAQ rich results and increases visibility in People Also Ask
QAPageSingle-question pages with several answersHighlights top answers in search results
HowToStep-by-step instructional contentEnables visual How-To snippets in Google results
PersonAuthor-led content (e.g. blogs, expert guides)Reinforces E-E-A-T by associating content with a real individual
OrganisationBusiness or brand-owned contentBuilds authority and legitimacy for your company

When using schema, remember that you can add them together. For example, you can use the HowTo schema with the Person schema. 

Also, when implementing a schema on your website, be sure to test it using Google’s Rich Results Test

Final Words

The future of SEO is question-first content. Because of this, you need to focus on reshaping your keyword-based content into question-based content. 

Right now, to kick things off, you need to perform a content audit. Look at the content you currently have and think about how you can phrase it into question-style content. Once you’ve done this, you can then go on to develop fresh content following the question-based methods. 

Written by Lawrence Hitches

Posted 9 May 2025

Lawrence an SEO professional and the General Manager of Australia’s Largest SEO Agency – StudioHawk; he’s been working in search for eight years, having started working with Bing Search to improve their algorithm. Then, jumping over to working on small, medium, and enterprise businesses with SEO tactics to reach more customers on search engines such as Google, he’s won the Young Search Professional of the Year from the Semrush Awards and Best Large SEO Agency at the Global Search Awards.

He’s now focused on educating those who want to learn about SEO with the techniques and tips he’s learned from experience and continuing to learn new tactics as search evolves.