How to show up in ChatGPT’s answers

Over the past year, we’ve been tracing the shifting landscape of SEO – trying to understand what’s changing, what isn’t, and how to keep showing up in a world where AI is now the first reader of almost everything. Mainly we’ve been trying to understand what patterns are emerging from these hidden structures.

Recently, that work has taken us from theory into practice. As we’ve begun experimenting with how AI engines find and summarise content, two new terms keep coming up: AEO and GEO.

AEO stands for Answer Engine Optimisation — the practice of making sure your brand or content appears inside the answers that large language models like ChatGPT, Gemini, and Perplexity generate. GEO stands for Generative Engine Optimisation — a broader version of the same idea, focused on how AI systems retrieve and summarise content across the web.

In essence, they describe the same thing: the next layer on top of SEO. Traditional Search Engine Optimisation was about ranking — getting your page to appear at the top of Google’s results. AEO/GEO is about being cited — showing up in the sources that generative models use when composing their responses.

We’ve learned that this new practice isn’t about replacing SEO; it’s about extending it. Your site still needs to be crawlable, structured, and authoritative. But now it also needs to be quotable — legible and to the machines that are writing the answers.

Traditional SEO was about ranking: owning a keyword, earning a link, being the destination. Answer Engine Optimisation (AEO) – or its sibling term, Generative Engine Optimisation (GEO) – is about being cited: appearing often enough, and credibly enough, that a language model includes you when it writes the answer.

In this new layer, authority isn’t concentrated in one URL – it’s distributed across the network of mentions, reviews, help-centre posts, and community threads that the model summarises.

When we started testing this, the biggest realisation was that visibility no longer lives on a single page. A Reddit thread, a YouTube walkthrough, or an authentic blog mention can weigh as much as a high-ranking article. The optimisation target has moved from “one link at the top” to “a pattern of relevance across the web.”

Before we go further, it’s helpful to understand what a ‘citation’ is in the context of GEO. A citation means a reference or inclusion of your content by a generative-AI system (such as ChatGPT, Claude or Gemini) in its generated answer.

More precisely:

• When the AI produces a response to a user query, it retrieves, synthesises or summarises information from various sources. A citation occurs when your content is among those sources and is attributed or referenced as part of the answer.

• In GEO, the visibility metric is not just “rank in search results” but “how often your content is cited (or used) in AI-answers”.

A large language model (LLM) has two halves:

1. The pre-trained model – or the ‘core model’. This is everything the LLM was trained on during its last update. It contains whatever data was used to give it enough understanding of the world its creators decided on. For example, the early 3.5 version of ChatGPT was famously trained on a large portion of Reddit.

2. The retrieval system (RAG) – This is an LLM’s external knowledge. It looks up live sources from the web, pulls the most relevant pages, and summarises them into an answer.

Together, they form the answer engine. The model can only cite what it can retrieve, which means findability now depends on being both technically legible and socially referenced.

That’s the subtle but crucial difference between SEO and AEO:

• SEO optimises for ranking in a list.

• AEO optimises for inclusion in a summary.

Both matter but they reward slightly different kinds of visibility.

Recently, we’ve noticed three patterns that reliably move the needle:

1. Structured depth wins. Pages that clearly answer a core question and the likely follow-ups (“how does it integrate?”, “what’s the cost?”, “who’s it for?”) surface more often in AI summaries.

2. Authentic citations count. Community mentions, credible affiliates, and product videos show up in the model’s retrieval results far more than we expected.

3. Consistency compounds. Once a source is trusted — accurate metadata, clear structure, no spam signals — it tends to be re-used across multiple answer engines.

These are small, repeatable actions that keep classic SEO disciplines useful while making content more “answer-ready.”

Here are key factors that influence whether your content is likely to be cited by a generative-AI system under a GEO strategy — and how you can track such citations.

Begin with a concise statement that directly answers the likely user-query, then support with details. This format aligns with how AI-models scan for “liftable” sentences.

• Use bullet points, tables or short lists to improve extractability

• Match sub‐headings to natural language questions (e.g., “What is X?”, “How do I do Y?”) so your content mirrors conversational queries.

Use concrete data, facts, statistics, clearly referenced sources. AI systems prefer content that can be corroborated.

• Include author and date metadata, show when content was last updated. Freshness helps.

• Use schema types such as FAQPage, HowTo, Article, Dataset etc., so machines understand the type of content.

• Ensure HTML is clean, headings make sense, content is not hidden behind tabs or heavy JavaScript that bots can’t easily parse.

• Your content (and your domain) should signal expertise, author credibility, unique perspective or original research to stand out.

• Narrowly focused pages that serve a specific intent (rather than a generic mega-page) tend to work better for citations.

• Keeping content current helps. If you publish the “definitive” page for a query but don’t revisit it, you may lose citation likelihood over time.

• Monitor where your content is cited and update it when you spot mis-attribution, stale stats or outdated claims.

Since the “citation” in GEO doesn’t always translate into a standard click or session, you’ll need custom metrics beyond traditional SEO.

• Citation frequency: how often your domain or page is referenced in AI-generated answers.

• Impression share in generative answers: share of your target queries where your content was cited vs alternatives.

• Citation position: whether your content appears as the first cited source, second etc. A higher position signals stronger authority.

• Brand lift / new branded queries: citations may seed brand awareness even without click-throughs. Track growth in branded search terms.

• Referral signalling from AI-platforms: Where possible, track visits coming via AI-tools (although this is often incomplete).

• Manual checks: query your key questions in major LLM platforms (e.g., ChatGPT, Gemini, Perplexity) and see whether your pages are cited and linked.

• Monitor server logs / bot traffic: Look for crawler user-agents associated with AI-systems (e.g., GPTBot, Google-Extended).

• Build a GEO scorecard: Select top 3-5 priority queries, track monthly whether you were cited, whether there’s a link, and what pages appear.

• Use mention-tracking tools: e.g., Brand24, BuzzSumo, specialised GEO tools may track citations by AI.

Search isn’t dying. It’s branching.
Every answer engine still begins with a search, but now it compresses the top dozen results into a narrative instead of a list. That means your work doesn’t just need to rank; it needs to teach the model what to say.

Or, put differently:

SEO helps people find you.
AEO helps machines explain you.

The best strategy, we’ve learned, is to design for both.