We know that Search is moving beyond the traditional results page. Rightly or wrongly, Google is slowly moving away from the traditional pack-based results based on what its algorithms have been looking over the past two decades or more.
The recent data proves that people are increasingly asking questions through Google AI Overviews and AI Mode, ChatGPT, MS Copilot and other generative AI platforms. Instead of presenting the typical ten links and asking the user to choose one, these systems are able to research any given subject, gather information from multiple sources and present a direct answer, all within the same search page. Most of the time, it completely negates the need to click through to a third-party website.
This new landscape creates many visibility challenges for brands and publishers. It’s also evolving, so it’s not easy to keep a guide such as this one up to date. This is essentially what we know at this point.
Traditional Search Engine Optimisation is primarily concerned with getting a page discovered, indexed and ranked. On the other hand, AI search adds another layer. Can an AI system even understand your information, trust it and select it as a source when constructing an answer?
This is where terms such as AEO, Answer Engine Optimisation, or sometimes termed as GEO, Generative Engine Optimisation, have recently emerged.
The terminology is still evolving, but the business objectives for an AEO consultant is straightforward:
- Increase the likelihood that your organisation, products, expertise and web content are represented accurately within AI-generated answers.
- Can AI-generated clicks assist with sales and conversion?
Is GEO really different from SEO?
Google’s position is that established SEO best practices remain in line with AI Overviews and AI Mode. In short, it means there are no additional technical requirements to appear in them. Pages still generally need to be indexed and made accessible to Google’s systems to be eligible for Search.
That is important, but by no means does AI optimisation end “conventional” SEO, in my opinion.
Meanwhile, OpenAI operates its own dedicated OAI SearchBot for content that may appear within ChatGPT Search.
As recently as last year, OpenAI has also been found out for looking at Google SERPs for resources. Reports as far back as August 2025 say that OpenAI scrapes Google, as well as some tests done by the SEO industry at large. So, with this in mind, we can’t discount the fact that AI search may dip into traditional search indexes, not all the time but occasionally, to discover content. With this in mind, traditional rankings through SEO can also be key.
Thinking Beyond Rankings
With traditional SEO, the primary questions for an SEO consultant were often:
- Where does my page rank?
- Is it (the page or site) attracting qualified organic traffic?
- Does the page satisfy the user’s search intent?
However, AI search introduces several additional questions:
- Is my brand mentioned?
- Is my content used as a source?
- Which page is cited?
- Which facts about my organisation are being repeated?
- Does the AI understand what our company, products and people actually do?
- Are competitors being recommended when we are not?
It seems to me that brand is playing a much more prominent role in AI search. This is supported in an Ahref analysis which found 75,000 brands that branded web mentions correlated more strongly with AI visibility than classic SEO metrics such as domain authority, referring domains and backlink counts.
While Google ranking can still be won by having an extremely strong page on a relatively unknown website (I know since I’ve had success). Page relevance, links, internal architecture, technical accessibility and intent matching can take that URL a long way in traditional search.
But AI answers often need another layer of confidence:
- Do multiple independent sources recognise your organisation/company/brand?
- What is it known for?
- How consistently is it described as being?
- Is it regularly associated with consistent topics?
This means your web visibility can no longer be measured entirely through rankings and organic clicks.
A potential AI visibility model is:
| Layer | Objective |
| Discoverability | Can search engines and AI retrieval systems find the page? |
| Understanding | Can they determine exactly what the page, company, product or person represents? |
| Extractability | Can individual answers, facts and claims be identified easily? |
| Authority | Is the source trustworthy to support an answer? |
| Validation | Do other independent sources support the your information? |
| Citation | Is the page actually selected as a source within AI responses? |
Traditional SEO remains fundamental to the first investigative layers. GEO becomes increasingly important as you move towards extraction, validation and AI citations. However, some underlying questions remain about sentiment. This means that does brand negativity and positivity affect mention/answer frequency for broader questions. At the moment, this is an unknown quantity.
The most valuable point to remember is the shift from keyword relevance towards contextual relevance and recommendation suitability.
Let’s get on to the guide.
Write for relevant questions, not simply relevant keywords
Keyword research does still remain useful, but AI conversations have to be more specific than search queries.
Someone might previously have searched:
“Enterprise CRM software”
An AI query may instead be:
“What CRM systems are suitable for a UK financial services company with more than 500 sales staff that needs Salesforce integration and strict data governance?”
Any content strategy therefore needs to move beyond isolated keywords towards questions, problem-solving, comparisons, criteria and decision-making scenarios.
Pages should anticipate the questions customers ask before, during and after a purchase.
Useful subjects include definitions, comparisons, advantages and disadvantages, suitability, costs, implementation issues, limitations, alternatives, compatibility, use cases and frequently encountered problems.
This creates content that can satisfy both conventional searches and the more detailed questions being asked through AI.
Build out knowledgeable topic depth rather than producing pages that say the same thing.
It was once an SEO trend to build out many pages about one topic, in order to become some kind of authority. Creating one page for every slight keyword variation is becoming increasingly difficult to justify.
The objective is to establish clear topical expertise. A stronger approach is to build authoritative resources around a subject. Plan different connected themes and increase internal linking throughout.
A software provider discussing data migration, for example, might cover migration planning, security, costs, timelines, common failures, platform compatibility and post-migration testing across a connected group of resources.
Don’t let backend AI systems run content generation on autopilot, allowing AI systems to run amok creating any kind of generic web pages through your gap analysis. It never ends well, so don’t do this.
Create content that AI needs to cite
One of the strongest opportunities for GEO is information gain.
We know that generic content is easy to reproduce to fill gaps. If twenty websites explain a topic in almost the same way, we can assume that AI system has very little reason to rely specifically on your version.
This does not mean every page should become a list of robotic FAQ responses. It means important information should be clear, self-contained, and logically organised.
Headings should accurately describe the subject beneath them. Definitions should be explicit. Product characteristics should be clearly stated. Use properly formatted tables where comparison information needs to be easier to understand.
Back in 2025, Microsoft recommended the use of clear headings, creating concise sections, Q&A formats, lists, comparison tables and supporting evidence to help make content easier for AI systems to interpret and reference. These essential practices should form part of your AI search optimisation strategy, although, that said, no clever formatting technique can guarantee inclusion in an AI-generated answer when your overall content does not match intent.
A page might contain a straightforward rewrite of what already exists and rank and garner low traffic. However, a page containing quality research and evidence nobody else has may become an important source. This quality difference is becoming more important for AI search.
If you must repeat information, at least publish additional material that’s meaningful to the existing body of information.
That could include options such as: original research to back up points, proprietary data, survey results, testing, expert commentary or opinion, user-generated content (when tightly moderated), original statistics, benchmarks, detailed case studies, methodology, photographs, diagrams or experience from real projects. We’ll discuss more about original data later.
Google itself increasingly emphasises unique, non-commodity information when discussing visibility in AI generative Search.
Make any of your claims specific and verifiable
AI systems need to distinguish between factual information and promotional language.
Statements such as:
“We provide market-leading solutions”
This statement offers very little information.
Compare this with:
“Our platform supports 38 currencies and integrates with Salesforce, HubSpot and Microsoft Dynamics through native connectors.”
The second statement contains identifiable entities and verifiable facts (perhaps from other resources).
Where appropriate, include dates, authors, methodology, data sources, product specifications and references.
If you publish research, explain how the research was conducted.
If an expert provides commentary, identify the expert and explain their relevant experience.
The objective is not to add artificial signals of authority. It is to make genuine expertise easier to evaluate.
Use images and video as information assets
AI search is increasingly multimodal.
Images should therefore contribute information rather than simply decorate the page. Use meaningful filenames, relevant alternative text and appropriate captions. Place images near the content they illustrate.
Try to utilise original photography, charts, product imagery, diagrams and annotated illustrations can provide information unavailable from generic stock imagery.
Any video should similarly be supported by clear titles, descriptions, transcripts and surrounding text.
Google reports multimodal search activity in Search Console and continues to expand the role of visual information within Search.
For some sectors, optimising visual assets may become as important as optimising paragraphs of copy.
Make the answer easy to extract
Here’s what we should be doing; What we should avoid is making the reader search through an introduction before reaching the answer.
AI systems frequently retrieve sections of pages rather than treating every page as one long piece of content.
This makes your document structure important. Use the classic heading hierarchy in the correct sequential order (H1 to H6), being as descriptive as possible.
A useful page should allow both humans and machines to understand each section without needing to interpret several paragraphs of marketing language first.
There is little evidence that says bullet points or numbered lists massively benefit AI search visibility, but they do, at least, help the reader pick out information when scanning.
By all means, use concise headings, semantic lists and comparison tables, but frame it as a way of improving presentation for the user. Imagine that you are making sure robot crawlers can access your important information in a well-formatted HTML page.
Build up authority away from your own website
One of the biggest differences between simplistic SEO thinking and a wider GEO strategy is the importance of information about your organisation elsewhere on the web.
A company can describe itself however it likes on its own website.That is one of the advantages of having one’s own website but this doesn’t necessarily give a clear, unbiased view of the brand or company.
Independent sources provide corroborating statements about your business that validate whatever you say.
Relevant signals might come from industry publications, professional associations, news coverage, research citations, expert interviews, conference appearances, reviews, supplier pages, partner websites and reputable directories. Each channel can provide varying degrees of validation.
This makes digital PR, thought leadership and expert contribution important parts of AI visibility.
The objective should not simply be to acquire another meaningless backlink.
The bigger objective is to establish a consistent external digital footprint that helps systems understand:
- Who the organisation is.
- What it is known for.
- Whether other credible sources recognise that expertise.
Brand mentions therefore matter even when they do not fit neatly into a traditional link building model.
Links remain valuable, but the wider relationship between brand, topic and independent sources deserves considerably more attention.
Own original statistics and research
Original data is particularly valuable because other websites have a reason to reference it.
Research can subsequently produce coverage, citations, backlinks, brand mentions and AI references.
For organisations capable of collecting meaningful data, this can become one of the strongest intersections between PR, SEO and GEO.
Instead of publishing another article entitled:
“Ten trends affecting the ETF industry”
a company might analyse several thousand customer interactions and publish:
“Stocks and Share Buying Behaviour Report 2027”
The second asset has much greater potential to become part of the information ecosystem that AI systems retrieve from.
GEO affects commercial pages, not just informational articles
The AI search discussion sometimes focus too heavily on editorial content.
Commercial pages also need to provide enough information to answer questions.
A product or service page should make its proposition clear.
- Who is it for?
- What problem does it solve?
- How is it different?
- What does it cost?
- What does it integrate with?
- Are there restrictions?
- What evidence supports the claims?
- What happens after purchase?
- Is your product or service validated elsewhere?
Pages containing little more than sales copy may give retrieval systems insufficient substance to confidently recommend the product.
This is particularly important because conversational AI allows users to ask increasingly specific product comparison questions.
Technical considerations
Strengthen entity understanding
AI search is highly dependent on understanding relationships between things.
Your organisation is an entity.
Your products are entities.
Your employees, locations, services, research reports and brands can also be entities.
Web content should consistently explain these relationships.
A company page should make it obvious what the organisation does, where it operates, who it serves and how its products or services relate to one another.
Product naming should remain consistent.
People should have useful author or profile pages where expertise genuinely matters.
Organisation, Person, Product, Article and other appropriate structured data can reinforce these relationships.
Regardless, if all AI search platforms place importance on structured data markup, it still needs to be there in your document. Google uses structured data to better understand page content, but markup should describe information genuinely present on the page rather than attempting to manufacture any new context or meaning for the product or service.Google also uses a scertain structured data to enhance the results pages.
All important content is available in the HTML
Your page needs to render enough content for systems to see.
Javascript-powered functionality can seriously hurt your page visibility when it comes to AI search.
A visually sophisticated webpage is not necessarily an easily retrievable webpage.
Critical information should be available to crawlers without depending unnecessarily on user interaction or complex client-side JavaScript.
This is particularly important for product information, category listings, specifications, FAQs, pricing information and core editorial content.
Where content is generated dynamically, organisations should review whether server-side rendering, static generation or another rendering strategy can make the primary information reliably available in the initial HTML.
AI optimisation does not remove traditional technical SEO requirements.You can read more about my look into JS content here.
Check which AI crawlers can access the website
Monitoring Crawler policy is just as important as technical SEO governance.
Blocking a crawler may prevent particular AI services from accessing page content. Be sure not to block access to the robots.txt file.
After much debate, ignore implementing the LLMs.txt file since Google, OpenAI, Anthropic, and Meta do not use it to control crawling. There are far better things to do technically than worry about a file that is not used. Stick with controlling the crawl by using the robots.txt
For example, OpenAI states that websites wanting their content included in ChatGPT Search summaries and snippets should permit OAI SearchBot. GPTBot is controlled separately, allowing organisations to make different decisions about search visibility and model training.
Organisations should therefore audit robots.txt, CDN rules, web application firewalls, bot protection systems and server responses.
Eliminate content duplication
AI retrieval does not make duplicate content problems disappear.
If several URLs contain almost identical information, systems still have to determine which version represents the subject.
Use consistent canonicalisation, internal linking, redirects and sitemap inclusion to reinforce the preferred version.
Microsoft specifically warns that duplication can make intent signals more difficult for AI systems to interpret and reduce clarity around which version should be surfaced.
Long-running large, international, ecommerce websites should pay particular attention to this area. It is advised to run segmented site crawls for products and any news sections to weed out these duplicate issues.
Maintain accurate XML sitemaps
Sitemaps remain useful because AI search still depends heavily on web discovery and search infrastructure.
Microsoft recommends combining XML sitemaps with IndexNow to improve discovery and freshness across traditional and AI-powered search experiences.
This becomes particularly important for frequently changing information such as products, availability, prices, news, jobs and events.
Publishing good information is not enough if retrieval systems discover the change weeks later.
Again, look out for removing duplication and unnecessary page references to increase crawl efficiencies.
Improve internal knowledge relationships by internal linking
Internal linking remains important, but the objective should be broader than distributing PageRank.
Links help establish relationships between topics.
A strong article about pension consolidation should naturally connect with relevant pension products, retirement planning guidance, tax information and supporting research.
Descriptive anchor text helps provide additional context about those relationships.
A well-structured site therefore begins to resemble a knowledge network rather than a collection of isolated landing pages.
This benefits users, search engines and AI retrieval systems simultaneously.
Monitor AI visibility separately from traditional SEO
Today’s visibility monitoring is slowly combining some traditional metrics with AI-specific data. It’s not perfect, but in time AI search systems and major search engines will roll out improvements.
Useful measures include organic rankings, traffic and conversions alongside AI citations, cited URLs, referral traffic from AI platforms, brand mentions, competitor inclusion and the types of questions where the brand appears.
Microsoft already exposes citations, grounding queries, topics, intents and citation share within Bing Webmaster Tools.
Google (within Search Console and GA) has also introduced reporting for appearances within generative AI Search features, including information about pages and countries where content appears, as well as the all-important clicks from citations.
OpenAI attaches the GA standard ‘utm_source=chatgpt.com’ parameter to their source referrals from ChatGPT Search, providing another useful form of analytics tracking.
The reporting model therefore needs to expand from:
rankings → clicks → conversions
towards:
visibility → citation → brand exposure → visits → assisted influence → conversion
For example, reporting 200% traffic growth needs context. Rapid traffic growth from a relatively small starting point doesn’t necessarily mean AI is delivering substantial traffic or revenue for every industry.
Measuring AI referral traffic alone is insufficient. AI can influence decisions without generating a website click, so it’s worth monitoring brand mentions, citations, recommendation frequency, and accuracy.
The biggest takeaway for marketing executives: SEO, GEO and CRO should no longer be treated as entirely separate activities. Businesses need to make their content understandable and credible enough for AI to recommend, then ensure their websites convert the visitors who arrive.
- Audit AI visibility: Test whether the likes of ChatGPT, Perplexity, Gemini and other AI platforms mention your brand for relevant questions, comparisons and use cases.
- Measure AI referrals: Identify which AI platforms send the most visitors and compare their performance against traditional acquisition channels.
- Use your analytics to improve important landing pages: Add stronger product details, comparisons, FAQs, evidence and information that helps customers make decisions. Explore user heatmaps of your pages.
- Test conversion improvements: Identify one or multiple important customer journeys, remove any potential friction points and measure their impact.
Where should my organisation start?
Trying to optimise an entire website for AI at once is probably unnecessary. Concentrate initially on commercially important subjects/topics for your brand. Then focus on optimising key pages. Also look at your existing analytics tools to find out which pages perform and make a start from there.
Here’s a basic starting point for your investigation.
- Identify the questions customers ask around your most important products and services. 50 unique examples as a starting point.
- Check whether your organisation appears when those questions are asked in major AI platforms
- Identify which websites and competitors are currently being cited
- Improve the depth, clarity and originality of the relevant pages
- Add evidence, expertise, research and clear entity information
- Check crawler access, rendering, canonicalisation and indexing
- Strengthen external authority through PR, partnerships and expert coverage
- Measure citations, brand visibility, referrals and commercial impact
This creates a manageable approach that can be expanded once your organisation understands where genuine opportunities exist.
The strategic takeaway
The debate over whether GEO / AEO is simply SEO risks missing the more important issues.
Google is correct to say that the foundations of good SEO remain crucial. SEO is certainly not dead. Websites still need crawlable pages, useful content, strong internal architecture, trustworthy information and clear technical signals.
But the interface through which people discover information is changing.
Search engines are no longer simply deciding which pages to rank.
AI systems increasingly decide which sources to retrieve, which facts to extract, which organisations to mention and which information to combine into an answer. This makes any quick-win solutions much harder to exploit. That changes the optimisation problem. The understanding is keeping your content aimed squarely at solving and answering issues that concern your audience.
If you require more information, or even a free 30-minute consultation, please contact: dave@davidchungdigital.com