SEO, GEO, AEO and more: engine optimisation explained

What is engine optimisation?

Engine optimisation is the work of making your content findable, and then chosen, by whichever system someone uses to look something up. That system used to be a list of ten blue links. Today it is just as likely to be a written AI answer, a chatbot, or an app store search box.

Seven acronyms now compete for the job. Here is what each one means, precisely:

Term	Stands for	What it optimises for	A distinct discipline?
SEO	Search engine optimisation	Ranking pages in search results that people click	Yes, the base layer
GEO	Generative engine optimisation	Being quoted or cited inside AI-generated answers	Partly; it has research behind it
AEO	Answer engine optimisation	Being the direct answer (snippets, voice, AI answers)	No; overlaps with GEO
AIO	AI optimisation (sometimes "AI Overview optimisation")	Visibility in AI features, often Google's AI Overviews	No; a label for GEO
LLMO	Large language model optimisation	How chatbots like ChatGPT describe and cite you	No; a label for GEO
SXO	Search experience optimisation	What happens after the click: speed, clarity, conversion	Partly; SEO plus UX and CRO
ASO	App store optimisation	Ranking and converting in the Apple App Store and Google Play	Yes, separate rules

The honest summary: SEO and ASO are real, separate disciplines with documented rules. GEO is a real research idea. AEO, AIO and LLMO are mostly marketing labels for the same idea, used by agencies and tool vendors to name a service. No search engine or standards body defines them.

Where the terms came from

GEO is the only new term with a paper behind it. In November 2023, researchers from Princeton University and IIT Delhi published GEO: Generative Engine Optimization (Aggarwal et al.), later accepted at KDD 2024. They defined generative engines as systems that use large language models to gather and summarise information for a query, and GEO as helping content creators improve their visibility in those answers.

AEO is older. It grew up around featured snippets and voice assistants, where one answer is read out and the rest of the page is ignored. AIO and LLMO arrived with ChatGPT and AI Overviews, and the definitions vary by whoever is selling them. SXO joins SEO with user experience: it argues that ranking is wasted if the page then disappoints. ASO is search for apps, and runs on rules that Apple and Google publish.

Why does engine optimisation matter now?

Because more answers are now given on the results page itself, and fewer people click through. The numbers have moved fast.

Google's AI Overviews reached India, in English and Hindi, in August 2024, as All India Radio's news service reported. By May 2025, Sundar Pichai said in his I/O 2025 keynote that AI Overviews had over 1.5 billion users in 200 countries and territories. A year later, at I/O 2026 on 19 May 2026, he said AI Overviews had over 2.5 billion monthly active users and AI Mode had passed 1 billion monthly active users within about a year.

India matters in this story. Google brought AI Mode to India as an English-language Labs experiment on 24 June 2025, and said that in the US and India, AI Overviews had driven a more than 10% increase in usage for the types of queries where they appear. On 7 October 2025, Google said AI Mode would be available in over 200 countries and territories.

The cost to publishers is clicks. The Pew Research Center studied the browsing of 900 US adults in March 2025. Users clicked a traditional result in 8% of visits when an AI summary appeared, against 15% when it did not. They clicked a link inside the summary in just 1% of visits.

Being found is no longer enough. Your page now has to be the one an AI system trusts enough to quote.

How do AI Overviews and AI assistants choose their sources?

They retrieve pages from a search index, break those pages into passages, and quote the passages that answer the question most clearly and credibly. The details differ by engine, and each has documented part of it.

Google: AI Overviews and AI Mode

Google's AI features documentation (last updated 10 December 2025) says both features may use "query fan-out": they run several related searches across subtopics and data sources, then pick supporting pages as the answer forms. That is why AI answers often link to a wider set of sites than the top ten results.

The same page is blunt about requirements. There are "no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." A page must be indexed and eligible to show in Google Search with a snippet. Google lists no new AI text files, markup or schema. Sites cited in AI features are counted in the normal Search Console Performance report, under the "Web" search type. Google also says clicks from AI Overviews tend to be higher quality, with visitors more likely to spend time on the site.

Microsoft: Bing and Copilot

Microsoft has been more specific about format. In an October 2025 post, Bing product manager Krishna Madhavan explained that assistants like Copilot "break content down", a process called parsing, into smaller structured pieces that are ranked and assembled into answers. The post recommends descriptive headings, question-and-answer pairs, lists and tables, specific claims instead of vague ones, and self-contained answers that make sense out of context. It warns that content hidden in tabs, locked in PDFs or shown only in images may be skipped.

In February 2026, Bing added an AI Performance report to Bing Webmaster Tools in public preview. It shows how often your pages are cited in Copilot and Bing's AI summaries, which pages are cited, and sample "grounding queries" the AI used to retrieve them. In May 2026, Bing's team put the shift plainly: "Search optimizes for likelihood of relevance. Grounding must measure strength of evidence."

OpenAI: ChatGPT search

OpenAI's crawler documentation separates three bots. OAI-SearchBot decides whether your site can appear in ChatGPT search answers. GPTBot collects content that may train future models. ChatGPT-User fetches a page when a user asks. Each is controlled separately in robots.txt, so you can stay visible in ChatGPT search while opting out of training. OpenAI says robots.txt changes take about 24 hours to apply.

What the research says works

The GEO paper tested nine content changes on a benchmark of 10,000 queries across 25 domains. According to the full paper, the best results came from adding quotations, adding statistics and citing sources, with gains of roughly 30 to 40% on its main visibility measure. Keyword stuffing gave little to no improvement. Lower-ranked pages gained most: citing sources lifted visibility by 115.1% for pages ranked fifth in search results. The authors also found that what works varies by domain, so treat these as starting points, not laws.

How do you optimise for every engine at once?

You run one plan, not seven. The underlying work for SEO, GEO, AEO, AIO and LLMO is the same, SXO extends it past the click, and ASO runs in parallel if you have an app. This is the order I would follow.

1. Make the site crawlable and indexable

Every AI answer engine above draws on a search index. Check that robots.txt, your CDN and your host do not block Googlebot, Bingbot or OAI-SearchBot, that important pages are in your sitemap, and that key content is in HTML text, not images or PDFs. Google's AI features page lists exactly these basics.

2. Answer the question in the first sentence

Write each section so its first one or two sentences answer the heading on their own. That is what query fan-out and Bing's parsing reward: passages that make sense when lifted out of the page. Use question-style headings, short paragraphs, lists for steps and tables for comparisons.

3. Add evidence: sources, numbers, quotations

This is where the GEO research is most useful. Name your sources in the sentence, link them, and date your numbers. It also matches Google's helpful content guidance, which describes E-E-A-T (experience, expertise, authoritativeness and trustworthiness) and says trust matters most.

4. Show who wrote it and how

Google's guidance asks three questions of any page: who made it, how it was made, and why. Use a real byline linked to the author's background, and say plainly if AI helped. Google warns that using generative AI to produce lots of text without oversight "represents little to no effort".

5. Fix the experience after the click (SXO)

Google says its ranking systems "look to reward content that provides a good page experience", and its page experience documentation names Core Web Vitals as signals its ranking systems use. Make pages fast, readable on a phone, free of intrusive pop-ups, and clear about the next step. Since AI answers send fewer clicks, each click is worth more.

6. Optimise app store listings separately (ASO)

If you have an app, follow each store's own rules. Apple's App Store search guidance says results depend on text relevance (title, subtitle, keywords and primary category) and user behaviour such as downloads, ratings and reviews. The keyword field is limited to 100 characters, and promotional text does not affect ranking. Google Play's store listing best practices limit titles to 30 characters and short descriptions to 80, and ask for everyday language, not lists of keywords.

7. Measure citations as well as clicks

Track Google Search Console for clicks, Bing Webmaster Tools' AI Performance report for citations, and a fixed set of questions you ask ChatGPT, Gemini and Copilot each month to see who they name. Compare the pages that get cited with the ones that do not, and rewrite the second group.

When should you do each step?

Start with the foundations in the first month, then work in cycles. A plan I would run:

• Week one: a crawl and index check. Fix robots.txt, sitemaps, blocked bots and any content trapped in images or PDFs.
• Weeks two to four: rewrite your ten most important pages so every section answers first and every number has a linked source. Add bylines and a line on how the content was made.
• Month two: speed and mobile fixes, then ASO if you have an app.
• From month three: a monthly review of Search Console, Bing's AI Performance report and your question set, and one round of rewrites based on what is and is not cited.

Expect a lag. Google notes that recrawling after a change can take days to months, and OpenAI says robots.txt changes take about a day to reach ChatGPT search.

Where does engine optimisation go wrong?

It goes wrong when teams chase acronyms instead of answers. Four common mistakes:

• Buying separate GEO, AEO and LLMO services. These are largely the same work. Paying three suppliers for one job creates conflicting advice.
• Keyword stuffing for AI. The GEO paper found it gave little to no improvement, and on Perplexity it performed worse than doing nothing.
• Inventing special files and markup. Google says no new AI text files or schema are needed. Structured data helps only when it matches the visible text.
• Ignoring the click. A page that wins a citation but loads slowly or buries the answer wastes the rare visit it earns.

In India, there is a further gap. AI Overviews work in Hindi as well as English, yet much regional content is published as images, scanned PDFs or social posts that search engines cannot read well. Brands selling beyond the metros should publish their key answers as plain HTML text, in the languages their customers search in.

Who should own engine optimisation?

One owner, usually the SEO or content lead, with clear help from three other teams. Web developers own crawlability, speed and structured data. Writers and subject experts own the answers, evidence and bylines. Analytics owns the monthly citation and click report. If there is an app, the product or growth team owns ASO, because it depends on ratings, reviews and release notes as much as on keywords.

Small businesses do not need a specialist for each acronym. A founder or marketer who writes clear, sourced answers on a fast, crawlable site is already doing most of the work.

My view

My view is that the alphabet soup is mostly noise. I spent years managing performance budgets across Google, Meta and affiliate networks, and the lesson that carries over is simple: platforms change their surfaces often, but they keep rewarding the same things. Clear relevance, real evidence and a good experience after the click.

GEO is worth learning because it puts evidence at the centre, and that is a healthy correction for content written to please an algorithm. AEO, AIO and LLMO are useful names for a conversation with a client, not separate plans. SXO and ASO deserve their own attention because the work and the owners really are different.

The trade-off is real. Writing answer-first, sourced content takes longer than writing for keywords, and AI answers may still keep the reader on the results page. Some brands will see citations rise while clicks fall. I would accept that, measure both, and judge the work by enquiries and sales rather than traffic alone.

Start with your ten most important pages

Pick the ten pages that bring you the most enquiries. For each one, check that it is indexed, that each section opens with a direct answer, and that every number links to its source. Then ask ChatGPT, Gemini and Copilot the questions those pages answer, and note who they cite. That list is your engine optimisation plan.