LLM Visibility: What It Is and How to Optimize for It

If your brand isn’t appearing in these AI responses, you’re becoming invisible to an increasingly important segment of your audience.

LLM visibility is about making sure you’re mentioned and cited in large language models (LLMs) like ChatGPT, Claude, Perplexity, and Google’s AI Overviews and AI Mode.

Tracking LLM visibility is important because more and more people are using LLMs to ask questions, get recommendations, and do things for them.

For example, SparkToro found that over 20% of Americans are now heavy users of AI tools (employing them 10X or more each month), and nearly 40% use AI tools at least once a month.

Simply put: If you’re not appearing in LLM summaries or citations, you’re missing out on potential leads, sales, and conversions.

For Ahrefs, for example, AI search visitors convert at a 23x higher rate than traditional organic search visitors.

Search engine visibility is about ranking higher in search engines like Google. Generally speaking, the higher you rank, the more clicks you’ll get to your website.

SEOs and digital marketers measure this by looking at:

• How much search traffic they get
• What keywords they rank for
• What positions they’re ranking for

On the other hand, LLM visibility is about showing up in the generated responses or citations by LLMs.

Most of the time, there’s no click as there’s no need to. Although that doesn’t mean you’re not getting conversions or sales.

People might still remember your brand (if mentioned). Or they might belong to the tiny group of people who actually click on the citations and navigate to your website. Or they might see your brand in a generated response and do a search for you on Google.

In the last scenario, even though the ‘click’ originated from LLMs, your analytics tool might attribute it to search engines, giving you the false impression that LLMs do nothing for you.

Although LLM visibility is the new kid on the block, it doesn’t mean it’s more important or that search engine visibility is dead.

Firstly, people are still using traditional search engines. According to SparkToro, 95% of Americans continue to use them each month, and 86% are heavy users.

Secondly, because of how LLMs work (see next section), you’ll likely still need to rank high in search engines in order to be cited.

For example, we found a positive moderate correlation between ranking high on the SERPs versus being cited in AI Overviews (AIOs). This suggests that the higher you rank, the more likely you’ll appear in AIOs.

Benji Hyam of Grow&Convert, a content marketing agency, also found the same. When they analyzed 400+ high-intent keywords where their clients ranked on page one, they found that, on average, their clients were mentioned 67% of the time in ChatGPT and 77% of the time in Perplexity.

In fact, according to Dan Petrovic of Dejan Marketing, an SEO agency, OpenAI has likely made the executive decision to make GPT-5 more reliant on retrieving from search engines, rather than trying to contain all information.

Let me translate: SEO is not dead. Search engine visibility is still as important as ever.

But I don’t want to say that SE visibility is 100% LLM visibility. There are still some gaps. For example, AI Mode (using Gemini) uses query fan-out (QFO) to uncover subqueries and long-tail keywords, which eventually gets combined into one generated answer.

Which means: They might end up even selecting pages that would have typically ranked on the second or third page of Google for the primary keywords. Traditionally, these pages would never get any visibility, but now they might.

To sum up: Both search engine and LLM visibility are important. Think of it this way: SEO is your foundation and generative engine optimization (GEO) is future-proofing your online presence.

In order to optimize for LLM visibility, we need to go back to first principles.

How do LLMs get their information?

In essence, two main ways:

• They ingest new training data to update their knowledge — This happens infrequently. For example, GPT-5’s cut-off was September 2024 and Gemini 2.5 Pro was January 2025.
• They rely on search indices — LLMs use retrieval-augmented generation (RAG) to retrieve information from search indices like Google and Bing, then incorporate that information into their generated answers.

So, all LLM optimization tactics revolve around these two methods:

• Increasing your visibility in training data — If your brand is commonly mentioned in the data sources these LLMs use, then it’s likely your brand will appear where relevant. However, this may not be the most feasible in reality, as we have no way to know or control which data sources LLMs use to train their models
• Increasing your visibility in search indices — Aka SEO!

How do we do this? Here are some tactics that’ll help:

1. Build off-site mentions

This is probably the single most important thing you can do to improve your visibility in LLM outputs.

That’s because one way LLMs understand your brand and when it should recommend you, is by understanding how many other places in its dataset mention you and you in the correct context.

In fact, in our study of 75,000 brands to see which factors are most likely to influence brand mentions in AI Overviews, we found that brand web mentions showed the strongest correlation with AI Overview brand visibility.

So, if you can get your brand or products mentioned many times in many places in relevant context, you dramatically increase the likelihood that LLMs will recommend you on those topics.

While you’ll want your brand to be mentioned on every relevant page in the world, a good way to start is to focus and get your brand mentioned on the most commonly cited domains on the Internet.

Every LLM has their own preferences on the type of websites they tend to include, but there are strong commonalities amongst all of them.

For examples, they all seem to prioritize:

• User-generated content sites like Reddit and Quora
• Third-party review sites like G2 and CNET
• YouTube

You’d want to be present on these sites: make videos, participate in communities, and get your customers to leave good reviews for you.

Yes, if you’ve caught on by now, SEO/GEO/AEO is just good digital marketing.

Another way is to look at the most commonly cited domains for your desired topics and make sure you’re mentioned in those websites too.

Here’s how you can do this:

1. Go to Brand Radar
2. Enter your topic
3. Go to the Cited domains report

These websites are all great candidates for outreach—you can guest post, sponsor them, or any other way to get your brand talked about or mentioned on those sites.

2. Optimize LLM-preferred content

Some page types are more likely to be cited by LLMs. Understanding these preferences can help you prioritize which pages to optimize and what new content to create.

When we analyzed AI traffic received by 35,000 websites to find the most common N-grams in page URLs, we saw these commonly cited page types:

• Blog posts and guides
• Comparison content (e.g. “top”, “best”, “vs”) as LLMs often need to present multiple options to users
• Core website pages like contact us, about us, and product pages
• Original research, studies, and data as LLMs frequently seek evidence and research to back up their claims
• PDF documents
• Video content

What do you do after you’ve identified these pages on your website?

Well, LLM optimization is still evolving and it’s impossible to guarantee anything, these techniques do show some promise:

• Keep content fresh — Our analysis of 17 million citations found that ChatGPT, Copilot, Gemini, and Perplexity all prefer to cite content that’s newer than what typically appears in traditional search results. This likely reflects how RAG works as LLMs are more likely to retrieve information about topics not included in their original training data. Focus on updating pages that cover fresh, modern, or novel topics rather than updating everything constantly.
• Use “Bottom Line Up Front” (BLUF) — Lead with your most important ideas as quickly as possible. This helps both readers and LLMs understand your content’s value immediately.
• Write declarative sentences — LLMs prefer confident, definitive statements when gathering evidence. Instead of wishy-washy language, write in opinionated sentences that sound authoritative.
• Keep sentence structure simple — Don’t start a paragraph about a topic but wait until the end to actually name what you’re discussing. Simple, clear sentence structures are easier for both readers and LLMs to understand.
• Increase entity richness — LLMs seem to prefer content that’s “entity rich”, i.e. text that includes many related products, topics, and concepts packed together. Mention different related entities clearly and frequently throughout your writing.
• Provide global document context — For longer documents, periodically remind readers (and AI) what the document is about. Include contextual reminders about the article’s main topic, especially in lengthy PDFs or comprehensive guides.

3. Monitor hallucinated URLs

In our study of 16 million URLs cited by ChatGPT, Perplexity, Copilot, Gemini, Claude, and Mistral, we found that AI assistants send visitors to 404 pages 2.87x more often than Google Search.

Yes, LLMs hallucinate a lot. Even URLs. So, even if they’re sending traffic to your site, they could be sending visitors to a page that doesn’t exist.

For example, a common hallucinated URL for our site was ahrefs.com/keywords. We write about keywords a lot, so LLMs expect that we have a page like that. But we don’t. So, those clicks are ‘lost’.

You’d want to find these hallucinated URLs with repeated LLM visits and redirect them to actual pages.

Here’s how to find made-up URLs:

1. Go to Ahrefs’ Web Analytics
2. In the Pages block, click Exit pages and select Possible 404.

This will show you pages that receive visits but have a title containing “404″ or “not found,” suggesting they may not exist.

Look through the list and set up 301 redirects from the hallucinated URLs to the best-fit pages.

4. Optimize for novel training data

LLMs are also trained on data that didn’t impact SEO or that SEOs didn’t care about. Examples include:

• GitHub
• Wikipedia
• Research papers (e.g., arXiv, PubMed)
• Patents
• Books

While your goal is not to spam these sites, if they’re relevant to you in anyway, you’d want to make sure information about your brand on these sites is accurate and consistent. If it’s even possible, make sure it references your brand too (where it makes sense).

5. Plug your entity gaps

Generally speaking, LLMs mention your brand based on their understanding of your entity relationships.

The words used near your brand actually influence LLMs’ understanding of your brand. They’re known as co-mentions.

It’s possible for your brand to have a disconnect between the topics you’d like to be visible for versus the topics you’re actually visible for.

You can use Brand Radar to see your visibility for key topics. You’d want to look for outputs that reference your competitors, but not you, and decide if this is a context you want to be associated with.

For example, let’s say you’re fashion brand Aimé Leon Dore. Enter your brand into Brand Radar and your competitors in the “Your Competitors” section. Then, click on the specific LLM you want to analyze, hover over your brand, and choose “Others only”.

This will show you all the queries that feature your competitors, but not you. You’d want to look through these topics and see if you want to be associated with them.

If yes, we can plug these entity gaps by creating on-site and off-site content about those topics.

6. Avoid too much JavaScript

Most AI crawlers do not render JavaScript.

So, if you have important parts of your website that are entirely JavaScript-based, there’s a high chance that they’re actually invisible to LLMs and will not get cited or included in AI conversations.

Don’t panic though. AI crawlers are ramping up and we’ve seen huge increases in AI bot activity on the Internet.

So, do expect this to change at some point and LLMs will start to render JavaScript. But for now, avoid putting your important pages in JavaScript.

7. Avoid spam

Yes, AI has made content creation infinitely easier and cheaper.

For example, 87% of AI users reported a cost of $0–100 per blog post, compared to 39% of non-AI users.

It’s very tempting for any company to just spam content for every imaginable topic or keyword and hope that they get crawled, indexed, and referenced by LLMs.

But we caution against this for a few reasons.

Firstly, Google is much less susceptible to this type of gaming. They’ve had years of practice filtering out spam content like this. And Google is still by far the most dominant source of traffic for most websites.

Second, even if it “works” right now on LLMs like ChatGPT, it’s only a matter of time before they catch up with this level of sophistication. Especially if they’re going to make RAG their primary method of getting up-to-date information.

Finally, even if you’ve successfully spammed LLMs, what now? The ultimate goal of LLM visibility, like any other channel, is to earn customers.

You might have hit your KPIs internally as the marketing team, but spammy AI content will never get you new business or earn the goodwill, trust, and affinity from people.

To track your brand’s presence in LLMs, you’ll have to use a tool like Ahrefs’ Brand Radar.

Enter your brand and you’ll see all how many times you appear across different LLMs:

To see how you’re comparing against your competitors, enter their brands in the “Your competitors” section.

To track how many visitors you’re getting from LLMs or AI search, install Web Analytics on your website. You’ll be able to see the amount of traffic you’re getting from AI search:

What about llms.txt?

Don’t bother.

LLMs already use most of the same infrastructure search engines use to crawl and understand your content (e.g. robots.txt). So, there’s no need for another standard like llms.text.

There is simply no current evidence that shows that using llms.text improves AI retrieval, boosts traffic, or enhances model accuracy. And no single LLM has committed to parsing it.

Final thoughts

The question isn’t whether AI search will become mainstream—it already is. The question is whether your brand will be part of the conversation when it happens.

The good news is that many of the tactics for improving LLM visibility are built on solid SEO and digital marketing fundamentals.

So, even though LLM optimization is still evolving rapidly, we can still count on one core principle: build a great brand.