How do you write content that AI cites? — 7 practical techniques

Would you like ChatGPT to use your site as a source? Seven techniques, with examples, that you can use in your next piece of writing today. You don’t need to be number one.

See whether ChatGPT mentions your business — free test

How do I get cited by AI? Short answer: answer concrete questions directly in the first lines, use structured data (FAQPage and Article schema), make the content readable without JavaScript, and build third-party signals in directories and reviews. This guide shows the writing technique step by step.

AI crawlers don’t index your website because it looks nice — they index it because the content is written in a format they can use. Research behind GEO (Aggarwal et al., KDD ’24) shows that targeted text changes can boost visibility in AI responses by up to 40 per cent. Here are 7 specific techniques you can use today to make your content citable by ChatGPT, Perplexity and Gemini.

Why does the AI cite your competitor — and not your company?

AI citation (also known as AI citation bait) is the practice of writing content optimised to be cited by generative AI models. In 2026, more and more Danes are actively using AI assistants such as ChatGPT, Perplexity and Gemini to search for local services, products and recommendations. AI citation differs from traditional SEO in that it focuses on factual accuracy, definition-based answers and structured data — rather than keyword optimisation and backlinks. It is the part of GEO that is about being used as a source. Being recommended by name is a different outcome, and the difference is explained further down.

The KDD ’24 study revealed something important for small businesses: when content is optimised, lower-ranked pages benefit the most. A page ranked 5th gained up to +115% visibility by citing its sources — whilst the same change cost the already top-ranked page visibility. So you don’t need to be number 1 to be cited. A separate study of content structure (Yu et al., 2026 — note: preprint, not peer-reviewed) measured a +17.3 per cent increase in citation rate solely due to better structure, tested on 200 benchmark articles across six AI engines.

Technique 1: Start with a definition — always

AI models love clear definitions. When ChatGPT or Perplexity need to explain a topic, they look for a paragraph that begins with “[Topic] is [short definition]”. This is the easiest way to get cited.

How to do it: Every page on your website should have a 25–50-word definition paragraph at the top of the main text. Write as if you’re answering the question “What is X?” — because that’s exactly what the AI does.

Example (good): “GEO (Generative Engine Optimisation) is a method for optimising your business’s visibility in AI-powered search engines such as ChatGPT, Perplexity and Gemini. GEO differs from traditional SEO in that it optimises for AI responses rather than search results pages.”

We are not aware of any published data on how much of an impact a definition placed at the start of the text has — so we do not claim any specific figure. The reason is technical: an AI model must be able to extract an independent answer from your page without reading the rest. A paragraph beginning with ‘X is …’ can be quoted as it stands. A paragraph beginning with ‘Background’ cannot.

Technique 2: Write fact-rich content — 3 facts per 1,000 words

AI models value quantitative data over qualitative statements. In the KDD ’24 study, adding statistics ranked among the best group of methods, with an improvement in the range of 30–40 per cent on the study’s primary visibility metric.

Every time you make a claim, ask yourself: “Can I put a figure on it?” Instead of “We deliver quickly”, write “We respond to enquiries within 24 hours on weekdays”. One requirement: the figure must be something you have measured yourself and can stand by. A made-up figure with a source attached is worse than no figure at all — both your reader and the AI model can check it.

Minimum requirement: At least 3 facts per 1,000 words. Facts can be percentages, prices, dates, measurements or references to external sources — and the source must be a link that the reader can follow.

Technique 3: Use BLUF — Bottom Line Up Front

AI crawlers do not read the entire article before deciding whether to cite it. They scan the first 60 words. If the answer isn’t there, they move on to the next source.

The BLUF rule: The first 60 words on your page must contain a complete answer to the question: who is this for, what is the problem, and what is the solution? An AI must be able to extract a two-sentence answer from your first 60 words.

Bad example: “In this article, we’ll go through some of the key aspects of GEO optimisation and look at how your business can benefit from the new opportunities...”

Good example: “GEO (Generative Engine Optimisation) optimises your business’s visibility in ChatGPT and Perplexity. It differs from SEO in that it optimises for being quoted in the answer — not for ranking on a results page.”

BLUF is the cheapest of the seven techniques: it takes five minutes and requires no new research — you simply move the answer up. We are not aware of any published impact measurement for this specific change, so we do not put a figure on it.

Technique 4: Link to external sources — greatest effect if you’re not ranked No. 1

The most interesting single finding in KDD ’24 is that the effect of citing your sources depends on where you’re already ranked. For a page ranked No. 5, it resulted in a +115 per cent increase in visibility. For a page ranked 2nd, it yielded just +2.5 per cent — and for the page already at the top, the effect was negative (−30 per cent). Take this as good news if you’re the underdog: source citations are precisely where the challenger has the most to gain.

What works: Link to HBR articles, university research (such as KDD ’24), industry reports (Bain & Company, BrightLocal), or public datasets. Avoid linking to your competitors’ websites — link to neutral, third-party sources instead.

Practical tip: Whenever you mention a statistic or a study, add a source reference as a link. This signals to the AI model that your content is research-based and credible.

Technique 5: Structure your content using lists, tables and bullet points

KDD ’24 did not test lists and tables — the study focused on text modifications. However, the 2026 study on structure (Yu et al.) measured precisely that: better organisation of content resulted in a +17.3% citation rate across six AI engines. AI models can easily extract and reproduce bulleted lists in their responses.

Here’s how: Make active use of HTML lists (

and
) and tables () in your content. Every list or table is a potential AI citation. Ensure that each bullet point can stand alone as a meaningful response.

Example of a list format that AI frequently cites:

• Quotes from named individuals: the study’s best single method — 30–40% on the visibility metric, 15–30% on the subjective assessment (KDD ’24)
• Statistics: the same top group, 30–40 per cent (KDD ’24)
• External source references: +115 per cent for a page ranked 5th — but −30 per cent for a page already ranked 1st (KDD ’24)
• Better content structure: +17.3 per cent citation rate (Yu et al. 2026, preprint)
• Keyword stuffing: has a negative effect — 10% worse than the baseline in Perplexity (KDD '24)

Technique 6: Include expert quotes — with name, title and link

AI models such as Claude (via Brave Search) place enormous weight on E-E-A-T signals. A quote from a named expert with a LinkedIn profile and CVR number can make the difference between being cited or ignored.

It is the best-documented of all seven techniques. At KDD ’24, adding direct quotes was the strongest single method out of the nine tested — a 30–40 per cent improvement on the study’s primary visibility metric. Use a genuine quote from a named individual, not an anonymous “the company says”.

Here’s how to do it, SME-style: Write as your founder or owner. Use “Martin Nymann, Founder at Geoa, says:...” instead of “At Geoa, we believe...”. Include the LinkedIn profile and CVR number — these are the signals AI models look for when assessing credibility.

Technique 7: Update the content regularly — AI loves fresh dates

Freshness seems to matter — but we haven’t found a metric that’s been documented well enough for us to put a figure on it. So we won’t. Treat it as a reasonable assumption, not as a measured effect.

Practical workflow: Go through your most important pages every three months. Update statistics, add new sources, and update the date in the JSON-LD schema. It takes 15 minutes per page. One word of warning: only update the date when you’ve actually changed the content — a date that’s moved without any changes is a false signal, and that’s exactly the sort of thing that both readers and models learn to spot.

Which techniques worked — and by how much?

Three stood out, and they all involve incorporating verifiable material into the text: statistics, quotes and source references. The figures below are the study’s total score for Position-Adjusted Word Count, where the baseline without optimisation is 19.3.

Technique	What it is	Score (baseline 19.3)
Quotation Addition	Insert relevant quotations	27.2
Statistics Addition	Insert relevant statistics	25.2
Fluency Optimization	Make the text easier to read	24.7
Cite Sources	Cite reliable sources	24.6
Authoritative	A more authoritative tone	21.3
Keyword Stuffing	More keywords in the text	17.7

The best methods raised the baseline by 41% for one metric and 28% for the other. Note the order: the three techniques that work best all require you to provide something concrete — a figure, a quote, a source. It’s difficult to cheat your way through.

Why did the techniques help most if you’re in fifth place?

Because the effect isn’t the same for everyone. The study calls this the ‘equaliser effect’: the lower-ranked pages gained significantly, whilst the top-ranked page lost out. It’s the most surprising table in the paper, and it’s almost never reproduced.

Technique	Rank 1	Rank 3	Rank 5
Cite Sources	−30.3%	+20.4%	+115.1%
Quotation Addition	−22.9%	+3.5%	+99.7%
Statistics Addition	−20.6%	+8.1%	+97.9%

The authors’ own explanation: generative engines base their results on content, not on backlinks or domain age. Consequently, small publishers are not penalised for factors on which they could never compete in traditional search. They call this an opportunity to democratise the field — and for a SMEs, it is the most encouraging line in the entire paper.

But do bear in mind the downsides too. All three techniques reduced the visibility of the page that was ranked number 1. So if you’re already at the top of your niche, the advice isn’t the same as for someone ranked number five. This is an important nuance that “add statistics and get cited more” doesn’t capture.

What didn’t work?

Two things, and both are worth knowing because they’re recommended anyway. Keyword stuffing — cramming in more keywords — scored 17.7 compared to the baseline of 19.3. In other words, it performed worse than doing nothing. The authors state explicitly that techniques which work in classic SEO do not necessarily translate to generative engines.

A more authoritative tone also failed to yield any significant improvement. The paper puts it as follows: generative models are already reasonably robust against such stylistic features. This echoes a widely held piece of advice: writing in a more assertive manner does not make you more citable. Having substance to back it up does.

What do our own Danish figures show?

They point in the same direction on one point and in the opposite direction on another. We asked four AI engines 576 customer questions about 465 Danish companies. The pages cited as sources had a technical score of 77.9 compared with 72.7 for the others — a highly significant difference (p=0.000). (Only Perplexity of the four conducted its own search; ChatGPT, Gemini and Claude obtained their sources from a shared search plugin — so, in their case, ‘cited’ refers to the search layer in front of the model.) This confirms the basic premise: what is on the page and how it is structured influences whether it is used as a source.

However, we also found something the study does not measure: technical setup predicted nothing about whether the company was recommended by name (74.9 versus 74.9, p=0.988). Citation and recommendation are two distinct outcomes, and the GEO research focuses solely on the former. The entire analysis centres on what determines whether AI mentions your company.

And on one point, our data directly contradicts the standard GEO recommendation: llms.txt. Among the plumbing firms visible in search results, 6 per cent of those mentioned had the file, compared with 13 per cent of those not mentioned; among accountants, the figures were 10 per cent versus 16 per cent. The figures are small and do not prove that the file is harmful — but we cannot support the claim that it makes AI mention you. The background to this is our measurement of the prevalence of llms.txt in Denmark.

How do you measure whether it’s working?

You can’t improve what you don’t measure. Here are three ways to track your AI citation score:

• Free GEO-test: Get a baseline score for your current AI visibility in 60 seconds. Take the test here.
• Weekly GEO measurement: With Geoa, you get automatic measurement across ChatGPT, Perplexity and Gemini — including which searches mention your business.
• Manual test: Ask ChatGPT “What is [your industry] in [your town]?” and see if you’re mentioned. Do the same in Perplexity and Gemini.

If you’d like to see these seven techniques applied to a real website, read the Relark.dk case study: starting score 49.9 (F), AI crawlers blocked in robots.txt, content that was only loaded via JavaScript — and exactly what was changed. We do not publish an ‘after’ figure for AI visibility until we have measured it over a period we can stand by. See how Geoa works.

Frequently asked questions

How do I get cited by AI?
The best-documented approach is to include verifiable material in the text. In the KDD study, the three techniques – Quotation Addition (quotes), Statistics Addition (statistics) and Cite Sources (source references) – scored highest, and the best methods raised the baseline by 41 per cent on one of the study’s two metrics. Our own Danish figures point in the same direction: the pages that AI cited as sources had a technical score of 77.9 compared with 72.7 (p=0.000).

Is it true that GEO can boost visibility by 40 per cent?
Up to that, yes — but that is a maximum, not an average. The figure comes from Aggarwal et al. (KDD 2024), was measured on a benchmark of 10,000 English-language searches, and the authors themselves state that the effect varies considerably across domains. On Perplexity.ai, they achieved up to 37 per cent.

Why do these techniques work best for low-ranking pages?
The study refers to this as an ‘equaliser effect’. ‘Cite Sources’ increased visibility by 115.1 per cent for a page ranked 5th and reduced it by 30.3 per cent for the top ranking. The explanation is that generative engines base their response on the content itself rather than backlinks and domain age, so that smaller publishers are not penalised for factors they could never compete on.

Does cramming in more keywords help?
No. Keyword stuffing scored 17.7 compared to the baseline of 19.3 — in other words, worse than doing nothing. A more authoritative tone also failed to yield any significant improvement. Techniques from classic SEO do not automatically translate to generative engines.

Does that mean AI will then recommend my business?
No, and that’s an important distinction. The research is about being used as a SOURCE. In our own survey of 465 Danish companies, technical setup showed no measurable correlation with being RECOMMENDED by name (74.9 versus 74.9, p=0.988). It is perfectly possible to be the page that AI reads without being the company that AI points to.

Sources and caveats

All figures relating to GEO techniques are taken from Aggarwal, Murahari, Rajpurohit, Kalyan, Narasimhan & Deshpande, “GEO: Generative Engine Optimisation", KDD ’24 — Table 1 (absolute scores) and Table 2 (relative change per ranking position). The study dates from 2024 and was conducted using English-language content and the authors’ own benchmark, the GEO-bench (10,000 searches), plus Perplexity.ai. The effect varies between domains; the figures here are presented with the method and metric to which they belong, precisely because they are often cited out of context.

The Danish figures are Geoa’s own measurement, 26 July 2026: 576 AI responses concerning 465 Danish company websites; the sample consists of companies visible in search results. The methodology and full disclaimers are set out in the full analysis.

Get started — your 5-step content plan

You don’t need to implement all 7 techniques at once. Here’s a prioritised plan:

1. Week 1: Add a BLUF and definition to your 3 most important pages. Update the dates in the JSON-LD schema. (1 hour)
2. Week 2: Review your pages for factual density. Add at least 3 facts per 1,000 words. (2 hours)
3. Week 3: Add external source references — link to HBR, KDD ’24, or industry reports. (1 hour)
4. Week 4: Structure the content using lists and tables. Add expert quotes, citing the source. (2 hours)
5. Going forward: Measure your GEO-score every week and repeat the process. Set a calendar reminder to update content every 3 months. (30 mins/week)

Would you rather have it done for you? The GEOA AI SEO employee™ reviews your pages every week and fixes the technical issues that decide whether AI can read them and use them as a source. The price you start on is the price you keep. Start free — 14-day trial, no card required.

Get your free GEO Score — 60 seconds, no credit card required.

Get started for free