Your organisation has been producing content for years. Rigorous reports. Documented positions. Analyses your peers cite in their own work.

Yet, when a policymaker or a citizen types a question about your topic into ChatGPT, you are not there.

This is not a quality problem. It is a structure problem.

This guide explains what happens inside AI engines when they select their sources, why institutional content is structurally disadvantaged, and which concrete adjustments can fix that — without rewriting everything.

AI Does Not Replace Google. It Sits Alongside It.

Before drawing conclusions in either direction, look at what the numbers actually say.

The Digital 2026 report by We Are Social and Meltwater, published in April 2026, gives a more nuanced picture than the usual headlines about the "death of Google". More than 4 in 5 online adults now use AI in some form. Among active users of platforms like ChatGPT or Gemini, the primary reason for use is finding information — cited by 58.8% of them.

But search engines are still used twice as much as AI for researching organisations: 45.8% versus 22.1%. Google's web audience remains seven times larger than ChatGPT's.

What is changing is the direction. The use of search engines for this type of research has declined by more than 10% in eighteen months. This is not a sudden shift. It is a slow migration — and it is documented.

In Belgium, 66.9% of internet users aged 16 to 24 say they used ChatGPT in the past month. These are the future members of your associations, the future donors of your NGOs, the future collaborators of your institutions. They are looking for answers on your topics. Not necessarily on your website.

One additional figure worth noting: according to Ahrefs, the click-through rate for the first Google position drops by 58% when an AI answer appears in the results. Classic organic traffic is shrinking. Visibility in AI answers is becoming a priority — not in ten years. Now.

Being Used and Being Cited Are Not the Same Thing

Here is the figure that deserves attention: despite its massive adoption as a research tool, AI does not appear in the top 15 sources for discovering new organisations. Only 14.8% of people say they discover new entities via AI tools.

This is not contradictory. People use AI to deepen what they are already looking for — not to discover what they do not yet know exists.

Which means that if your organisation does not appear when someone searches your topic, you disappear precisely at the moment when intent is strongest. The user knows what they are looking for. They ask a specific question. And the answer cites your competitors, not you.

This is not a reputation problem. It is a citability problem.

The distinction matters because it changes the nature of the solution. Building reputation takes years. Improving citability requires structural adjustments — which this guide details in the following sections.

What AI Engines Look For — and What Your Content Gives Them

Google looks for pages that match a query. ChatGPT or Perplexity look for sentences that directly answer a question. These are two different logics.

To understand why, it helps to understand how AI engines work. They use a process called RAG — Retrieval-Augmented Generation. In practice, it happens in five stages.

First, the engine interprets the user's question — not by keywords, but by intent and concepts. Then it searches its index for semantically relevant documents. It ranks them according to relevance, authority, freshness, and structural quality. It generates an answer by synthesising the best sources. And it cites those that provided extractable, verifiable content.

That last point is where institutional content runs into difficulty.

A sixty-page report may contain the best available analysis on a topic. If that analysis is never formulated as a direct answer to an identifiable question, it will not be cited. Even if it is online. Even if it is public. Even if it is impeccable in substance.

Institutional content tends to start with context before answering, to organise information by theme rather than by question, to favour nuance over direct formulation. These are genuine qualities in a report written for peers. They are obstacles for an AI engine looking for an extractable answer in 150 words.

There is, however, a structural advantage to seize. The primary concern of AI users is unreliability — hallucinations, false sources, factual errors. Credible, well-sourced organisations whose content is verifiable have a real structural advantage — provided their content is accessible in the right format.

SEO and AEO: Two Disciplines, One Direction

Traditional SEO and AEO are not opposites. They reinforce each other — as long as you understand what distinguishes them.

SEO optimises pages for search engines: titles, keywords, backlinks, domain authority. AEO — Answer Engine Optimisation — optimises content to be cited by AI engines: question-and-answer structure, modularity, structured data, freshness.

The good news: content well structured for AEO is almost always better ranked on Google. Both disciplines share the same foundation — quality content, well organised, factual and sourced. AEO adds one layer: each section must be independently understandable and independently citable.

One figure illustrates this convergence well: 38% of citations in Google's AI answers come from pages already ranking in the top 10 organic results. Working on SEO remains relevant — but it is no longer sufficient. The remaining 62% come from sources Google did not necessarily rank first, but whose structure was optimal for answer extraction.

Five Concrete Adjustments — Not a Full Rebuild

This is not about rewriting your site. It is about making what you already have citable.

Reframe section headings as questions

This is the simplest adjustment — and one of the most effective. AI engines identify sections by their headings. A heading formulated as a question immediately signals that the section contains an extractable answer.

"Our actions in 2024" becomes "What did we accomplish in 2024?". "Regulatory context" becomes "What are the current obligations in this sector?". "Our approach" becomes "How do we work with our partners?".

The change is minimal. The effect on citability is real.

Add a summary block at the top of each key piece of content

Three to five points condensing the essentials of the page or article. This is what AI engines read first — and what they most often extract for their answers.

This block must directly answer the main question of the content in the first 100 words. No preliminary context-setting. No "in this article, we will explore". A direct answer, immediately.

Create or expand a FAQ page

Not the questions you would like your audiences to ask. The questions they actually type into Google — and now into ChatGPT. "What is the concrete impact of your organisation?" "How do you use the funds you receive?" "What is your position on this issue?"

This is the format most directly citable by AI engines — because it corresponds exactly to the structure they are looking for: a question, a direct answer, verifiable information.

Include citable data at regular intervals

AI engines favour content that includes statistics, percentages, and precise figures with their sources. Content without a single data point is less citable than content that grounds each argument in measurable reality.

The practical rule: one sourced data point every 150 to 200 words. Not generic data — precise, dated, attributed to its source. "According to the Digital 2026 report by We Are Social" is infinitely stronger than "according to recent studies".

Implement schema markup

Schema markup is a technical language invisible to site visitors, but decisive for AI engines. It explicitly signals the nature of your content: article, FAQ, organisation, event.

Two types are particularly important for AEO: Article schema (or BlogPosting), which signals that this is dated, attributable editorial content; and FAQPage schema, which marks each question-answer pair as extractable. Both are added in your CMS page settings — in Webflow, this is directly accessible in the Page Settings panel.

The Mistakes That Prevent You from Being Cited

Knowing best practices is not enough if structural errors neutralise them.

Publishing content without sources. An unsourced claim cannot be cited by an AI engine that is specifically trying to avoid hallucinations. "Digital accessibility is improving" will not be cited. "According to the 2024 digital accessibility barometer, 40% of audited sites show little or no improvement effort" will be cited.

Leaving key analysis locked in PDFs. AI engines index web pages much better than PDF documents. An annual report of a hundred pages likely contains ten citable insights. None of them will be extracted if they remain locked in a PDF. The solution: transform key conclusions into structured web pages or articles. The PDF remains available for those who want the full document.

Neglecting content freshness. An Ahrefs study of 17 million AI citations found that pages cited by AI engines are on average 25.7% more recent than those cited by traditional search. Content that has not been updated in two years gradually loses its citability. Statistics age. Regulatory references evolve. An annual update of key articles is the minimum.

Structuring for experts, not for questions. Institutional content is often written for peers who share the same frame of reference. It begins with context, develops nuance, and arrives at the conclusion at the end of the document. This is exactly the opposite of what an AI engine is looking for — which wants the answer first, the development second.

Ignoring schema markup. Many organisations skip this step because it seems technical. It is an avoidable mistake. FAQPage schema alone can significantly increase a page's visibility in AI answers. It does not require rewriting the content — only marking it up correctly.

How to Measure Your AI Visibility

What is not measured cannot be improved. AEO is no exception — but its metrics differ from classic SEO.

The indicators that matter

AI citation count is the central metric: how often is your content cited by ChatGPT, Perplexity, Google AI Overviews, or other platforms in response to questions related to your domain?

Traffic from AI platforms is measurable in Google Analytics 4 by filtering traffic sources: chat.openai.com, perplexity.ai, gemini.google.com. This traffic is still marginal for most organisations — but it is growing at approximately 1% per month according to a study covering 3.3 billion sessions. More significantly, visitors arriving via an AI citation convert at 4.4 times the rate of classic organic visitors.

Share of voice in AI measures how frequently your organisation is cited compared to competitors on your core topics. It is the most strategic indicator — and the hardest to measure manually.

How to build a baseline

The simplest method to start: identify ten to twenty questions that your stakeholders frequently ask about your topics. Type them directly into ChatGPT, Perplexity, and Google. Document whether your organisation is cited, and if so, which content is cited. Repeat the exercise monthly.

This is a manual method, but it produces a real picture of your current positioning — and the questions on which you do not yet appear.

Available tools

For organisations that want to automate this tracking, platforms like Frase, Semrush, or Ahrefs have developed AI visibility tracking features. They allow you to monitor citations across multiple platforms simultaneously and identify content that is performing — or declining.

For those starting without investment, Google Analytics 4 is sufficient to measure incoming traffic from AI platforms. Google Search Console remains essential for tracking the impact of AI Overviews on organic traffic.

What Changes in the Next 18 Months

AEO is a young discipline. What works today will evolve — but the fundamentals will remain stable.

Google AI Mode will expand. Launched in 2025, Google AI Mode offers a fully conversational search experience within Google. Its progressive rollout means that optimising for AI answers within Google itself is becoming as important as optimising for classic results.

Multimodal search will grow. AI engines are increasingly processing images, video, and audio alongside text. Visual content accompanied by precise descriptions, detailed text alternatives, and transcripts will be favoured.

Voice and AI are converging. With 170 million voice assistant users expected in the US by 2028, AI-powered voice assistants will draw from the same sources as text-based AI engines. Conversational content structured in question-and-answer format will perform across both modes.

The first-mover advantage is closing. Today, 70% of organisations believe AEO will significantly impact their digital strategy within 1 to 3 years — but only 20% have started implementing it. This window is closing. Organisations that structure their content now are building authority that competitors will struggle to catch up with.

What will not change: quality, structure, and credibility remain the foundations of visibility — on Google, in AI answers, and on whatever engine exists in ten years.

Key Takeaways

What is AI visibility for an organisation? The ability of content to be extracted and cited by AI engines like ChatGPT, Gemini, or Perplexity in response to a user's question. It is not identical to Google ranking — but the two reinforce each other.

Why is institutional content rarely cited by AI? Because it is structured to inform expert readers, not to answer specific questions. AI engines extract direct, modular, clearly formulated answers. Institutional reports rarely meet this structural criterion.

Does the entire site need to be rewritten? No. Five adjustments are enough to start: reframe headings as questions, add summary blocks, create a structured FAQ page, include citable data, implement schema markup.

Is Google disappearing in favour of AI? No — not in the near future. But the use of search engines for researching organisations has declined by more than 10% in eighteen months. Both channels coexist. Best practices apply to both.

How do you measure AI visibility? By manually testing target questions on ChatGPT and Perplexity monthly, tracking incoming traffic from AI platforms in Google Analytics 4, and using specialised tracking tools to automate monitoring.