A practical guide to AEO: what I’ve learned from researching and testing it

If you have started hearing about Answer Engine Optimisation, or AEO, you are probably trying to work out what it actually means for your website and marketing.

I was in exactly the same position a few months ago.

There was plenty of information about “ranking in ChatGPT”, AI search, GEO, LLM optimisation and new AEO tools. But the more I researched it, the harder it became to separate the genuinely useful advice from the hype.

So I started breaking it down.

What is actually changing? What stays the same? What do we need to do differently? And, most importantly, can we measure whether any of it works?

Since then, I have spent hours researching AEO, attending webinars, reading technical guidance, speaking with people working in the space and trying to understand what actually matters. 

We have also been testing what we learn on The Lead Agency website and, now that we have a much clearer idea of what we are doing, applying the same thinking to client websites.

That practical testing has been just as important as the research. It has helped us work out which ideas are useful, which are probably being overhyped, and where we are actually seeing measurable changes.

AEO is still developing, and I am sure our approach will keep changing with it. But the basics are much easier to understand than they first appear.

So this is my attempt to explain the most important things I have learned in simple terms, including what we have tested ourselves and what has actually made a difference.

The short version

If you only remember a few things, I would make them these:

• SEO is still the foundation of AEO.
• Keywords still matter, but AI prompts give us much more context about what buyers actually want.
• Good AEO content should answer real questions clearly.
• Experience, proof and specific examples are becoming increasingly important.
• Your website needs to be accessible to search engines and AI crawlers.
• You need to track AI visibility if you want to know whether your work is having an impact.
• You probably do not need hundreds of new pages. In most cases, you need to make the content you already have more useful.
• There is no single “AEO trick”. It is a combination of good SEO, useful content, technical foundations and a better understanding of the buyer.

What is AEO?

Answer Engine Optimisation is about improving the chances that your business or content is included when someone asks an AI platform a question.

That could mean being mentioned, cited or recommended in platforms such as ChatGPT, Gemini and Perplexity, or appearing within Google’s AI Overviews and AI Mode.

The easiest way I have found to understand the difference is to compare it with traditional SEO.

Someone using Google might search:

B2B marketing agency Melbourne

Someone using an AI platform might ask:

We are a manufacturing company with an internal marketing manager but need help with SEO and Google Ads. Which B2B marketing agencies in Melbourne should we consider?

The first is a keyword.

The second contains much more context about who the buyer is, what they need and what they are trying to decide.

That difference looks simple, but it has probably had the biggest impact on the way I now think about search.

Does AEO replace SEO?

No. This was one of the first things I learned.

When I started researching AEO, I expected there to be a completely new set of rules.

There are definitely some new considerations, but the more I looked into it, the more obvious it became that good SEO is still doing a lot of the heavy lifting, at least for now.

The things we were already focusing on for SEO still matter:

• useful, authoritative service pages
• a clear website structure
• strong internal linking
• topical authority
• crawlability
• quality content
• credible proof.

In fact, our AEO research reinforced several SEO changes we were already making.

For example, we have been moving away from creating separate pages for every slight keyword variation. Instead, we are building stronger pages around broader topics and supporting them with related content.

This means that rather than having three weak pages competing with each other for similar terms, we would generally prefer one strong page supported by relevant blogs, FAQs, case studies and internal links.

One of the unexpected outcomes of researching AEO is that it has actually made us rethink parts of our traditional SEO process.

We are now asking more questions about topic coverage, search intent, how pages relate to each other and whether a piece of content genuinely adds something useful. Those are good AEO questions, but they are also good SEO questions.

AEO should build on good SEO, not distract you from it.

What is the difference between keywords and prompts?

Prompt research has changed the way I think about search more than almost anything else I have learned about AEO.

Keyword research tells us what people search for.

Prompt research can tell us much more about why they are searching, the situation they are in and what they need help deciding.

At first, I thought prompts were basically the AEO version of keywords. They are not.

A prompt can give you far more information about the buyer. It can tell you who they are, what problem they have, what they have already tried, what matters to them and the decision they are trying to make.

That was a genuine aha moment for me: prompt research is really customer research.

That means AEO research needs to start with a better understanding of the buyer.

I would not recommend opening ChatGPT and asking it to generate AI prompts without giving it any real customer context.

Before we develop prompts, we go much deeper into understanding the business and the buyer.

For every AEO project, we start with a 16-page knowledge pack that we work through with the client. It covers things such as:

• who the buyer is
• the problem they are trying to solve
• what has triggered them to start searching
• what they care about when choosing a provider
• the questions they are likely to ask at different stages
• what makes the business different
• the expertise, experience and proof we can draw on
• how the buying decision is actually made.

That level of detail matters because useful prompts need to reflect real buying situations, not generic questions generated by AI.

It has also changed the way we work with our clients. We are spending much more time getting their input before we write anything.

The deeper I have gone into AEO, the more convinced I am that we cannot create genuinely useful content without that subject-matter insight. We know marketing, SEO and content. Our clients know their customers, their industry and the questions they are asked every day.

The strongest content comes from combining both.

We then use that information to build prompts around different stages of the buying process, such as problem-aware, solution-aware, comparison and high-intent commercial.

For example, a problem-aware prompt might be:

Why is our website getting traffic but not generating quality B2B leads?

A high-intent commercial prompt might be:

Which Australian B2B marketing agencies can manage SEO, Google Ads and content for a business without an internal marketing team?

Those are very different questions and should not necessarily be answered by the same piece of content.

This is probably one of the most useful things AEO has taught me.

It has pushed us to think beyond “what keyword should this page target?” and ask “what is this person actually trying to work out?”

That is a much better starting point for content.

Do you need to write special content for AI?

Not in the sense of writing completely separate, unnatural content for AI.

There is a lot of advice around AEO telling businesses to add question headings, FAQs, summaries, tables, schema and short answers everywhere.

Some of that is useful.

But I do not think the answer is to make every webpage look like it has been written for a robot.

Instead, I would ask:

Can a person, or an AI system, quickly understand the answer this page is giving?

When we review content for AEO, I now look at six fairly simple questions:

• Does it answer the question directly?
• Can we support what we are saying?
• Is it easy to understand?
• Is the information structured clearly?
• Is it obvious who and what the page is about?
• Is the information still current?

This can mean using a direct answer underneath a heading, adding a table when it genuinely makes a comparison easier, or including an FAQ section where there are useful questions that have not already been answered.

But I would not add these things simply because someone said they were good for AEO.

The page still has to be useful for the person reading it.

Another thing I assumed at the start was that AEO would mean creating a lot more content.

In practice, that has not always been the case.

Sometimes the right page already exists. The issue is that it does not answer the question clearly enough, the information is too generic, important detail is missing, or there is not enough proof behind what is being said.

So before creating something new, I would ask whether you can make an existing page better.

Why does first-hand experience matter for AEO?

The more AEO content I researched, the more obvious one problem became: generic content is becoming less valuable.

AI is already very good at summarising information that exists online.

If you publish another article that says exactly the same thing as the other ten articles on the topic, there is not much reason for Google, ChatGPT or the person researching the subject to choose yours.

What is much harder to copy is your actual experience.

That might include:

• what you have done
• what you learned
• your own data
• client examples
• case studies
• results
• mistakes
• specific observations from doing the work.

This is also why the client input I mentioned earlier has become so important. That 16-page knowledge pack is not just a briefing document. It is where a lot of the material that makes the content genuinely useful comes from.

We know SEO, AEO and content. Our clients know their customers, their industry, the questions they are asked and what actually happens in practice. We need both.

AI has made that collaboration more important, not less. It is easy to generate generic content quickly. The tool is not the differentiator. The source material is.

This is something we are putting much more emphasis on in our own content.

And this article is a good example.

I could write a generic article explaining ten things businesses should do for AEO. But I think it is much more useful to explain what I have actually learned from researching it, testing it and seeing what happens.

For me, this is also where businesses with genuine expertise have an advantage.

AI can reproduce a generic explanation very easily. It cannot reproduce your project experience, customer insight, internal data or the conclusions you have reached from actually doing the work.

That means case studies, real examples, client insights and original observations are becoming much more important. They are not just there to support the content. Increasingly, they are the part that makes the content worth reading and potentially worth citing.

Is AEO only about content?

No, and this took me a little longer to appreciate.

When I started looking at AEO, most of my attention was on content.

Eventually, I realised there was another fairly obvious question:

Can the AI platform actually access the information?

This led to one of the things that genuinely blew my mind when I first learned about it.

If important content depends on JavaScript to render, you cannot assume every AI crawler will access it in the same way a browser does.

To a person visiting the website, the page can look completely normal. But if key content depends on JavaScript to render, a crawler may be seeing a much more limited version of the page.

The below example highlights the gap between what is visible to an AI bot compared with what a human sees when browsing this particular website.

Source: Humanitix

This was a good reminder that AEO is not just about what you write. You also need to think about what AI systems can actually access.

I had always thought about JavaScript mainly from a website speed or SEO perspective. I had not really considered that an AI crawler and a person could effectively be seeing different amounts of information.

That led us further into the technical side of AEO.

We started looking more closely at areas such as:

• AI crawler access
• robots.txt
• indexability
• JavaScript
• internal linking
• website structure
• schema.

You do not need to become a developer to understand the basic principle.

Your important information needs to be accessible, clearly structured and easy for search engines and AI systems to interpret.

JavaScript itself is not automatically a problem. The issue is that different crawlers have different capabilities, so you should not assume that every AI crawler will access dynamically rendered content in exactly the same way.

If important information depends on scripts, interactions or other rendering processes, it is worth checking what the crawler can actually access.

Schema is part of that picture too.

We use appropriate structured data to help search engines understand things such as the organisation behind a website, its services, articles, authors and how pages relate to one another.

I would not describe schema as a shortcut to appearing in ChatGPT. It is simply one of the technical signals that can help machines understand a website more clearly.

How do you know whether AEO is working?

This was one of the biggest challenges when we started.

We already knew some people were finding The Lead Agency through ChatGPT because they told us.

That was encouraging, but it was hardly a measurement strategy.

We could not answer some very basic questions:

• Which prompts are we appearing for?
• How often are we being mentioned?
• Which competitors appear instead of us?
• Is our visibility getting better or worse?

We started using Klove to establish a baseline and monitor our AEO performance.

That allowed us to track things such as prompt visibility, competitor performance, brand mentions and AI crawler activity.

Measurement was where AEO stopped feeling theoretical for me.

Before that, you could occasionally check ChatGPT, see your brand appear and assume you were doing reasonably well. Once we had actual data, we could see where we appeared, where we were missing and which competitors were appearing instead.

Instead of simply saying, “we need to optimise for ChatGPT”, we could look at where we were visible, where we were missing and what might need to change.

For example, if we were not appearing for an important commercial prompt, we could review whether we actually had a strong page answering that question.

If not, that gave us something practical to work on.

We could also see who was appearing instead of us, which gave us something specific to investigate rather than guessing.

What content do they have? What questions are they answering? Are they clearer about a particular service or topic? Do they have stronger proof?

Instead of guessing what AEO work we should do next, the data could start informing it.

So, does AEO actually work?

The real test, of course, was whether any of this actually changed our visibility.

You can follow every AEO recommendation available, but does it actually change what happens in AI results?

In our case, yes.

We established a baseline for The Lead Agency and then worked consistently on our content, website structure, prompt coverage and technical AEO.

After around three months, we saw a significant improvement.

In July 2026 compared with June 2026:

• ChatGPT visibility increased by approximately 150%
• Gemini visibility more than doubled
• Google AI Overviews visibility tripled
• AI bot visits increased by 238%
• our overall AI visibility score increased by 133%.

The Lead Agency also became the most visible brand across the 144 AI prompts we were monitoring. 

You can read the full AEO case study here to see what we changed and how our AI visibility improved. 

Source: Klove.ai

I do not think that means we have discovered a secret formula for “ranking in ChatGPT”.

I am also very conscious that the platforms and technology will continue changing.

What the results do tell me is that AI visibility is something we can monitor, work on and improve, rather than simply hoping an AI platform happens to mention us.

The other important point is that there was no one change that suddenly produced the result.

We worked on content, SEO, prompt coverage, internal linking, technical accessibility and measurement together.

That is why I am wary of advice that reduces AEO to one tactic, whether that is adding FAQs, implementing schema or creating an llms.txt file.

Individual technical changes may have a role, but none of them is an AEO strategy on its own. The improvements we have seen have come from working on content, SEO, prompt coverage, internal linking, technical accessibility and measurement together.

So, where should you start with AEO?

If someone asked me today where to begin, I would keep it fairly simple.

1. Get your SEO foundations right.
   Make sure the website is technically sound and that your important services have strong, useful pages.
2. Understand your buyers.
   Think about what they would actually ask an AI tool at different stages of their buying journey. Do not start with hundreds of AI-generated prompts. Start with real customer problems, questions and buying situations.
3. Check whether your website answers those questions.
   You may already have the right pages, but they might need more depth, clearer answers or stronger proof. Look at what you can improve before assuming you need another page.
4. Add experience and evidence.
   Use case studies, examples, results, reviews and your own knowledge rather than relying on generic information.
5. Check technical accessibility.
   Make sure your important information can be accessed and understood by search engines and AI crawlers. Pay particular attention to content that depends on JavaScript or user interaction to appear.
6. Establish a baseline and start tracking.
   Without measurement, it is very difficult to know whether anything you are doing is improving your AI visibility. Then use what you learn to decide what to work on next.

This is essentially the process we have been developing at The Lead Agency.

What is the main thing I have learned about AEO?

The more I have worked on AEO, the clearer it has become that it should not be treated as a completely separate discipline from SEO and broader digital marketing.

There are new platforms, new search behaviours and new things we need to measure, but the underlying principles are familiar.

Understand your customers, create genuinely useful content, demonstrate real experience, make that information easy to find and understand, and then measure what happens.

The biggest change for me has been how these pieces now connect. Prompt research improves our understanding of the buyer. AEO content principles are making our SEO content stronger. Technical AEO has made us think more carefully about what crawlers can actually access. And AI visibility tracking gives us another source of data to guide what we do next.

AEO is really an extension of good SEO and good marketing, not a replacement for either.

That is the framework I would use if I were starting with AEO today.

Want to improve your visibility in AI search?

If you are trying to work out what AEO means for your business, we can help you understand where you are currently visible, where the gaps are and what is worth working on.

Our approach combines SEO, content, technical AEO, prompt research and ongoing measurement, so the focus is not on chasing individual tactics, but improving your visibility across both traditional and AI-driven search.

Learn more about our AEO services.