15 Ways to Get Your Brand Cited by ChatGPT, Perplexity, or Google AI Overviews

SEO & GEO

15 Ways to Get Your Brand Cited by ChatGPT, Perplexity, or Google AI Overviews

Getting cited by AI engines like ChatGPT, Perplexity, and Google AI Overviews requires a strategic shift in how brands structure and distribute their content, a shift that sits at the center of GEO services as its own discipline. We asked marketers and SEO specialists who have successfully earned AI citations for their companies to share what actually worked, and added our own answer into the mix. Here are 15 tactics straight from the people who tested them.

1. Put the Company Name in Every Claim

2. Published Verifiable Client Results Upfront

3. Featured One Precise Internal Statistic

4. Moved Key Facts Into Static HTML

5. Unblocked AI Crawlers at the CDN Edge

6. Addressed Specific Buyer Concerns Directly

7. Refreshed Existing Pages for Current Relevance

8. Wrote With a Position Instead of Generic Advice

9. Rebuilt a Guide Around User Questions

10. Built a GDPR Page for ICP Queries

11. Turned HARO Into Source Distribution

12. Earned Third-Party Expert Attributions

13. Targeted Cited Sources With Outbound Outreach

14. Matched Executive Queries With Verifiable Facts

15. Built Interconnected Topic Clusters

Put the Company Name in Every Claim

One change did more than everything else we tried, and it’s a writing habit rather than a technical fix.

We stopped writing sentences that begin with “we.” Any sentence carrying a fact worth quoting now carries the company name inside it: not in the page title, not in a heading, but in the sentence itself.

The reason is mechanical. These systems quote sentences and short passages, not whole pages. If a sentence says “we tested this across client accounts and found the following,” the model can lift the finding and drop the source, because the source isn’t in the words it took. If the sentence says “Visionary Marketing tested this across client accounts and found the following,” the name travels with the fact.

I noticed this by accident. I asked one of the assistants a question we had published a clear answer to, and got our own finding back, phrased close to how we’d written it, credited to nobody. Our wording had made us anonymous.

We rewrote the key sentences on our main pages and asked clients to do the same on theirs. No new content, no schema work. Mentions of our name in the answers we check rose by about 35% over the following quarter.

Being the source isn’t the same as being named as the source, and only one of those is worth anything.

Christopher Coussons, Director, Visionary Marketing

Published Verifiable Client Results Upfront

For our own agency site, the thing that got us cited most consistently in AI Overviews and in ChatGPT answers about SEO agencies in Morocco was publishing a genuinely specific data point instead of a general claim.

Most agency websites say something like “we help clients rank higher.” We instead published a page stating an exact result: 12 of 15 client campaigns we ran over 18 months increased organic traffic by 40% or more within the first six months, with the methodology and date range attached. That single page (a clear number in the H1, supporting data in a table right below it) started showing up as a cited source in AI Overviews for queries like “SEO agency results Morocco” within about ten weeks of publishing.

What I think made the difference isn’t the number itself, it’s the format. Large language models pull from pages that state a claim and immediately back it with a verifiable, specific figure in the same paragraph, not three scrolls down. We also added a short methodology note explaining how we counted the 12 of 15, which campaigns were excluded and why, and I think that transparency is part of what makes a page trustworthy enough to cite.

Since then, we’ve applied the same structure across client sites: replace vague benefit statements with one real, sourced number stated plainly, placed as close to the top of the page as the content allows. Citation rate in AI answers for those pages has been noticeably higher than for pages we haven’t updated yet, though we’re still building a larger sample before I’d call it a hard rule.

RHILLANE Ayoub, CEO, RHILLANE Marketing Digital

Featured One Precise Internal Statistic

We published one narrow page built entirely around a single specific stat instead of trying to cover an entire topic broadly, which is honestly what most of our existing content had been doing up until that point.

AI tools seem to pull concrete, specific numbers far more easily than they pull general summaries. Broad coverage apparently gives them nothing specific enough to grab onto and cite back to a source. We took one number straight from our own internal data and phrased it as a clean, standalone line near the very top of the page, with nothing else competing for attention around it. It started showing up in AI Overview citations within a few weeks of publishing, something none of our longer, more comprehensive guides had ever managed, despite ranking perfectly fine in regular search.

Faizan Khan, PR and Content Marketing Specialist, Ubuy Singapore

Moved Key Facts Into Static HTML

We get direct citations from ChatGPT by moving key brand facts and pricing out of JavaScript widgets and into static HTML. It verifies things by probing sites directly, and it can only see text. If pricing information is hidden behind toggles or a bunch of scripts, it can’t see it, and it’ll cite a third-party review directory instead of the brand.

We simulate this by browsing client websites with JavaScript off, which reveals what the AI crawlers see. Positive reviews that get pulled up dynamically aren’t seen by the models, and neither are product specs. You have to put all the key numbers directly into the body text of the website.

Load times and text length matter too. The AI crawler typically gives up on a webpage after about 2 seconds and 2,000 words, so we put the most important information near the top.

Ulf Lonegren, Executive Director of AI, Sōvyn

Unblocked AI Crawlers at the CDN Edge

Our answer pages were fine: the CDN was returning a 403 to GPTBot and PerplexityBot, and had been for about a year. After eight years running organic search for a workwear brand, unblocking them at the edge was the single thing that made the difference. Nothing cites a page it can’t fetch.

We then rebuilt the sizing guide as one question per heading with the answer in the first forty words underneath, and put the fit data we’d collected from returns into a plain table with the sample size at the top. We now appear in 41 of a 200-prompt set we track monthly, up from three in March.

Check bot access before writing anything. Most of this is a plumbing problem wearing a content strategy costume. Perplexity picks us up; ChatGPT barely does. Our best-cited page is a table nobody in marketing wanted to publish.

Fahad Khan, Digital Marketing Manager, Ubuy Kuwait

Addressed Specific Buyer Concerns Directly

The biggest shift for us was writing content that directly answers specific buyer questions instead of general topic pages. We started building pages and blog posts around exact questions customers ask us in sales conversations (things like how enamel pin pricing works, or what the difference is between die-struck and soft enamel) and answering them clearly and simply near the top of the page. That style of direct, structured answer seems to be exactly what AI tools pull from when someone asks a similar question, and we’ve started showing up in AI-generated answers for a handful of those specific product questions.

The other piece was tightening up our internal linking so those answer pages connect clearly back to the actual product pages, since that structure seems to help AI tools understand what we actually sell and trust the page enough to cite it.

Eric Turney, President / Sales and Marketing Director, The Monterey Company

Refreshed Existing Pages for Current Relevance

The thing that made the difference was updating old content with sharper relevance instead of chasing more content. I reviewed established pages, removed generic lines, added fresher context, and reframed explanations around the questions people actually ask today. AI systems often favor pages that feel both experienced and current, especially when the information is concise and confidently expressed.

The site already had strong topical consistency and a clear connection to high-stakes business decisions, which gave those updates more weight. After the refresh, pages were easier to parse, easier to summarize, and easier to cite. In many cases, improving what already exists creates more AI visibility than publishing something new.

Brian Hansen, President, Rocket Pilots

Wrote With a Position Instead of Generic Advice

Most agency content sounds like it was written to rank, not to be believed. It hedges. It says “it depends.” AI models have no reason to cite a sentence that refuses to commit to anything, because there’s nothing in it worth attributing to a source.

We write from a position instead. On our own site, we don’t say agencies vary in how transparent they are. We say most agencies keep clients dependent on purpose, and that we built Tabula so clients own everything we build for them, whether they stay with us or not. That’s a specific, falsifiable claim tied to our name, not a hedge.

The mechanism is simple once you see it. AI systems are compressing the internet’s advice into the median opinion. A page that states the median opinion has nothing to stand out with, so it gets folded into the summary and stripped of its source. A page that stakes out a real position is the thing the summary has to attribute, because it’s the outlier, not the average.

My rule for anything we publish now: if a competitor could publish the same sentence unchanged, we haven’t said anything yet. Rewrite it until only we could have written it.

Being quotable and being agreeable are different skills. AI citation rewards the first one, not the second.

Carlos Rios, Founder, Tabula

Rebuilt a Guide Around User Questions

I took the single longest guide on my site (the one people had been linking to for years) and rebuilt it around the questions people actually type. Every subhead became a question in plain language, and the answer sat in the first two sentences under it, before any setup or story.

The second change mattered more. A lot of the useful detail was locked inside screenshots, charts, and step-by-step images, so it existed for a human reader and nowhere else. I rewrote all of that as plain sentences in the body copy: numbers, sequences, caveats, and exceptions, all stated in text. I also added who wrote it, when it was last reviewed, and what the guide doesn’t cover.

Then I tested it the boring way: I asked the assistants the same question five or six different ways, one phrasing at a time, and watched whether my page came back as a source. Where it didn’t, I read what got cited instead and found the specific sub-question my page never answered directly, then added a short section that answered it. Now I write with the question as the header, the answer in the opening lines, the specifics in text rather than in an image, and no brand plug anywhere near the answer.

Will Mitchell, Founder, StartupBros

Built a GDPR Page for ICP Queries

I built a dedicated GDPR page for Oh Dear from scratch, targeting the specific queries their ideal customers were searching for. That turned an existing, uncited mention into actual citations. Google AI Overviews now cites the page for queries like “GDPR-compliant uptime monitoring hosted in the EU for public sector websites.”

Deian Isac, Founder, goBOFU

Turned HARO Into Source Distribution

The thing I did was stop treating HARO as a backlink tactic and start treating it as source distribution. We built a repeatable process for getting my expertise, name, and company into genuinely useful third-party articles, which has resulted in more than 350 expert quotes and bylined pieces published. Those pages give AI search systems something stronger than a claim on your own website: an independent source connecting your brand to a specific topic. My rule for AI visibility is simple: don’t just publish more about yourself, become the source other publishers use when they explain the subject.

Callum Gracie, Founder, Otto Media

Earned Third-Party Expert Attributions

We stopped treating our own website as the only venue for our expertise and started pitching quotes to industry publications and media outlets. AI tools don’t pull from brand websites. They pull from the sources they were trained to trust: publications, expert roundups, third-party aggregators. Once GavelGrow started appearing as a named attribution in articles about law firm marketing and digital advertising, it started showing up in AI-generated responses to those same questions. Own-site content is nearly invisible to these systems; third-party attribution is the actual mechanism.

Abram Ninoyan, Founder & Senior Performance Marketer, GavelGrow

Targeted Cited Sources With Outbound Outreach

I scraped the URLs of the sources that were already being cited and launched a standalone outbound campaign to them, offering value-add data or resources those specific sites would want to reference. Adding them to the campaign’s target list tripled our visibility. Once a few of them linked back to or cited our content, ChatGPT and Perplexity started pulling us into the same answer sets as the sources we’d targeted.

Max Desiak, B2B SaaS SEO & AEO Consultant, Max Desiak Consulting

Matched Executive Queries With Verifiable Facts

One concrete move that worked for me: writing content in direct question-and-answer format, matching the exact phrasing people type into ChatGPT or Perplexity, not SEO keyword strings. When I published pieces on why 95% of AI projects fail in companies, I titled each section as a literal question executives ask, with the answer in the first two sentences. That structure gets pulled into AI Overviews because the model doesn’t have to infer intent.

Second: consistency of verifiable facts across independent sources. My Zumo Imagination exit to Huli Health, my Trifecta Software founding date, and my PROMIDAT ML certification all appear the same way across my site, my Primera Linea columns, and SUMMA’s coverage. LLMs weight corroboration across independent domains higher than a single polished bio page.

Third: I avoided generic AI-hype language entirely. Specific, falsifiable claims get cited more often than vague thought leadership.

Joe Phillips, Founder, Spearhead Technologies

Built Interconnected Topic Clusters

We found that strong topical continuity across pages helped us earn more citations. We didn’t want isolated articles that answered one question and then faded away, so we built connected groups of content where related topics supported each other from different angles. We kept the language clear and consistent so readers and AI systems could understand the relationship between each topic.

That consistency helped us build trust around the subject. When AI models see the same ideas explained clearly across several pages, they can treat the source as more dependable. We also made sure each page addressed one clear intent and linked naturally to the next step. This approach gave us more than visibility. It helped our brand become easier to recognize as a clear, reliable source.

Sahil Kakkar, CEO / Founder, RankWatch

The Takeaway

None of this is one trick. It’s the same shift showing up fifteen different ways: be specific instead of vague, take a position instead of hedging, and make the fact easy for a machine to lift and attribute. Pick two or three that fit how your team already publishes, and start there. It’s the same discipline our SEO services team applies to every page we write for clients.