The way people source information is changing. Alongside Google, many now ask ChatGPT, Perplexity, Copilot and other AI models directly for the answers they seek.
That presents a new challenge for organisations. These tools selectively cite sources to collect the relevant information and present a single, curated response. The latest thinking is that if your content isn’t used to inform that response, then in many ways it doesn’t exist.
Thus, we have seen the emergence of Generative Engine Optimisation (GEO), the younger sibling to Search Engine Optimisation (SEO). Where SEO achieves visibility through high rankings across different keywords in search engines, GEO works to include your content as a source for AI-generated answers.
Earning these citations could increase your brand’s visibility, establishing you as a trusted source of information. It could be a determining advantage against competitor articles that fail to make the cut and see a decline in performance and value.
Where to begin though? What should you be looking to accomplish with new content to obtain these elusive spots? We’ve investigated GEO and identified five strategies that will help determine whether your brand appears in AI-generated responses or risks falling behind.
1. Build voices of authority
AI models examine verifiable data to determine what’s credible. They prioritise content written by people with relevant credentials, referenced by trusted third-party sources and published on domains that other authoritative sites link to regularly.
In practice, this means curating your selection of author bylines and corresponding profiles based on their expertise.. Moreover, you should strive to publish accurate, up-to-date content that reputable industry webpages may want to cite and link to.
The challenge for most businesses is that authority builds slowly. A single well-researched article won’t immediately establish your organisation as a reliable source of information for AI models. However, consistently publishing expert-backed articles will build that credibility over time.
2. Structure content for easy extraction
As with SEO, ease of navigation and appropriate metadata hygiene will enable greater success. Prioritise content that any AI tool can parse quickly and confidently, including clear headings, properly formatted lists and direct answers. You should also include schema markup that identifies whether you’ve published an article, a guide or a product page.
Think and write in short segments rather than lengthy prose. Keep paragraphs short (two to three sentences), stick to one idea per subheading and only surface definitions or key figures in bullets, and avoid burying answers in long sentences. This reduces ambiguity for the AI system, while also helping your readers quickly locate the information they need.’
3. Pack pages with verifiable information
AI systems also favour content that delivers specific and relevant information in every paragraph. They are designed to avoid vague statements, such as “the solution improves efficiency”, and prefer claims backed by evidence, such as “reduces processing time by 40% based on a study of 200 enterprises.”
This requires some discipline from the content team. For informative articles, they should eliminate promotional language that lacks concrete backing and support claims with data, examples and references to credible sources.
Moreover, as with SEO, you should still answer your audience’s most frequently asked questions, using tools like Semrush or Ahrefs to analyse search volume. One well-researched, fact-dense article on a narrow but popular topic will likely outperform multiple lightweight pieces covering a broad area.
4. Build content that works across all platforms
Each AI platform operates differently. Perplexity cites sources by default, ChatGPT displays sources through its search feature and Google’s AI Overviews link to indexed pages. But chasing platform-specific optimisation is largely futile given how rapidly these tools are evolving.
Instead, focus on fundamentals that work universally: clear structure, attributed facts, current information and thorough answers. Content built on these principles performs well across the different models, regardless of their individual quirks. That, in turn, helps maximise the value you receive.
5. Check where you appear and then adapt
Your WordPress dashboard – and most CMS backends – won’t tell you when an AI answer cites your content. The most reliable way to check is manually: run your customers’ most frequently asked questions through ChatGPT, Perplexity and Gemini. Note whether your brand appears, which pages are linked and how they’re described.
Do this regularly, and you may start to notice patterns. Perhaps your how-to guides are frequently cited, but your thought leadership content is not. You may appear for certain topics but not others.
Tools are emerging to track AI citations, but they’re still in the process of maturing. Meanwhile, AI models are constantly updated, so what works today might not work in three months. Regular manual checking remains your best option for now.
Content that AI likes to cite
There are emerging instances of organisations that are doing GEO well, leading to increased citations within AI answers.
For example, Mayo Clinic’s pages that discuss various medical conditions typically open with a plain-English definition, then proceed through symptoms, causes and treatment in predictable sections, accompanied by medical bylines and review dates. That has led to frequent citations for Google’s AI overviews.
It’s similar for financial queries. Investopedia’s pages that define standard terms and acronyms begin with a one-line definition, followed by the formula (where applicable), a worked example and common pitfalls.
That ‘define, demonstrate and qualify’ pattern is easy for both readers and models to consume, which has led to GEO success: Investopedia recently ranked within the 100 most-cited domains on ChatGPT in both US and global datasets.
—
The most striking thing about GEO is how little it really asks you to change. A good content strategy has always prioritised expertise, clarity and credibility. What’s different now is that AI systems enforce these standards more rigorously than human readers.
Vague claims that busy professionals might overlook get filtered out by AI models seeking verifiable facts. Promotional language becomes an obstacle to machine parsing.
In that sense, these tools are less forgiving but more predictable. You just need to focus on the fundamentals of authority, clear structure and factual density, and your content will perform well across both traditional search and AI discovery. Ignore them, and you may disappear from both.
Some background reading
- GEO: Generative Engine Optimization (Princeton Study) – The original academic research from Princeton, Georgia Tech, Allen Institute for AI and IIT Delhi that established GEO as a field and demonstrated visibility can increase by up to 40%
- The 10 Most Cited Domains in AI Assistants (Ahrefs) – Analysis of 78.6M AI interactions showing which websites are cited most frequently
- 100 Most Cited Domains in ChatGPT (Ahrefs) – Data on which domains appear most in ChatGPT responses
- 76% of AI Overview Citations Pull From Top 10 Pages (Ahrefs) – Research showing the relationship between traditional SEO rankings and AI citations
- What is Generative Engine Optimization? (Search Engine Land) – Comprehensive overview of GEO tactics and best practices
By Will Walton
