AI-powered search has fundamentally changed how people find information online. Instead of going through search results, many people now ask LLMs or rely on the AI summary Google provides.
Naturally, website owners are starting to wonder if there are ways to make their website appear in more AI searches. A viewer brought up this exact topic during our March 19 2026, Office Hours livestream.
And while there’s no official playbook yet, we’ll explore a few good practices to increase your site’s chances of appearing in AI search results.
The viewer’s question
To begin, let’s take a look at exactly what our viewer asked. It’s a multi-layered question, so we’ll break it down further. The viewer is an agency owner, looking for information on behalf of their client.
A client is asking for their website to come up more in AI searches. I've submitted llms.txt, and I've advised them about the tips mentioned in the webinar by Daphne. Do you have a source for tweaking the llms.txt manually to improve this? Also, is there a way to track analytics from AI searches to the website?
Firstly, the webinar that the viewer refers to is SEO for LLMs: It's just SEO (mostly), in which Daphne Monro, our Head of Website and Content, discusses precisely this topic. We’ll go over a lot of what she says in this blog, but we still recommend you give it a watch as she offers a lot of great examples and stats.
Next, we’ll look at the llms.txt file. In short, it’s a file that holds a set of rules that LLMs may or may not abide by.
Finally, yes, there are ways to track traffic from AI search, but only partially. There are dedicated tools for that, but as Daphne puts it:
There are dedicated tools for tracking traffic from AI searches. However, we should be wary as each user’s experience with LLMs varies vastly.
Quality content still works best
We wouldn’t blame you if you thought that AI would have some specific SEO requirements. It’s a new technology, so of course, it must have some new hoops to jump through to appease it for the sake of SEO.
The good news is that’s not necessarily the case. The fundamentals of SEO haven’t changed. Nathan, our Office Hours host and agency coach, explained it like this:
The core things we’ve been doing for SEO don’t change. It’s still about quality content.
LLMs don’t rank pages like search engines do. They don’t have all the requirements and metrics that Google has, for example. However, like Google, they still rely on clear, well-written content that’s relevant to the question being asked and useful to real people.
In other words, if you are already writing content that ranks well in Google, you don’t need to change your approach for LLMs drastically. The only things to look out for are EEAT/HCU and content freshness (how recently something was updated). But, even then, both of those are still focused on helpful content, just from a different perspective.
Because of that, both Google and LLMs are great at identifying content that’s meant to answer, not merely rank. Content that:
Explains things clearly and concisely.
Breaks down complex ideas into understandable pieces.
Gets to the point quickly, without a lengthy preamble.
Think of it like this: your content should genuinely help people understand a topic, regardless of their level of knowledge. Depending on the audience you target, you might need to boil things down to manageable chunks of information or be very technical and detailed.
In either case, you are helping others further their understanding, and that’s what LLMs (and Google) prefer.
Make your content easy to understand
Strong content is a solid foundation. Making it easy for humans and AI to process will make it even more solid.
Think of your content as something that might get summarized. Whether it’s by a person or, more likely, Google’s AI overview or any of the other LLMs out there, we’ve all seen how AIs answer our questions.
This is why content structure and how easy it is to skim matter. Here are a few things to keep in mind when writing your pages, blog posts, or any other type of content.
Use clear headings: A clear, logical heading structure helps AI, people, and even crawlers better understand what your content is about.
Always have only one H1.
Each new section should be under an H2.
Use H3s for subsections directly relevant to the H2 topic.
Short paragraphs: Walls of text put people off fast. Using short, clear language is much more appealing to the reader.
Bullet points and lists: Imagine if we had splayed this list out in several paragraphs? It would be unsightly and difficult to follow. Instead, use a list to concisely deliver key points, instructions, or—as you might have guessed it—listing items.
Lead with answers: Don’t bury the answer to the question you are asking under paragraphs of text. Explain early, briefly and precisely; elaborate later. People who want the quick answer will be satisfied, and those who want to learn more will be, too.
FAQ sections and schema markup: An FAQ section is a great way to answer questions that didn’t come up in your post or page. Adding schema markup will also help AI understand the context of your content.
As you can see, LLMs don’t require you to write in any specific way. Not yet, at least, and we certainly hope that doesn’t change. In fact, there are studies that show LLMs struggle or even ignore structured data entirely, even when it’s provided to them.
Ultimately, LLMs digest content differently from humans; they don’t necessarily read a page. Instead, they process text as tokens (which are often limited), turning long content into chunks. They then summarize and gain understanding that way.
Because of that, structure, clarity, and skimmability are just as vital as it is when a human is reading your content. The fewer tokens an AI has to use to understand your content, the better it will do so.
So, if your visitors are happy with the clarity of your content, AI will be, too, even if for different reasons.
Build authority
Another good way to help your website be more noticeable by LLMs is to build up its authority. This, much like what we’ve discussed so far, will help with SEO in general, not just when it comes to AI.
The goal here is to make your website an authoritative source of information. A place that other websites can cite and refer to with confidence that your information is correct.
There is no single, easy-win way, but these tips will help.
Don’t focus on individual keywords; instead, cover entire subjects in depth. Doesn’t need to be in one article or page, as you can instead set up topic clusters that you can link to individual pieces of.
Speaking of, use internal links to strengthen your content. That will help people, search engines, and AI crawlers discover more of your website.
Use original research and data wherever possible. If you have an expert on hand you can quote, or data you can show, that will help build your site’s credibility.
Getting mentioned on third-party platforms (like Reddit) or collaborating authoritative websites will also help to strengthen your content quality. Brand mentions and guest posting are great for link building.
Authority is a long-term project that will need dedicated time and effort to achieve. Being highly regarded in the online world as a place where one can find good information is important nowadays. And, as you may have already guessed, it will help people trust you more, and search engines rank you higher.
What about llms.txt?
So far, we’ve looked at ways to make your website more enticing for LLMs, but our viewer mentioned something else in their question: llms.txt.
If you are familiar with the inspiration for llms.txt, robots.txt, then you already know what it is about. It’s not a direct copy of robots.txt, though, as their purposes differ.
Llms.txt is a file that directs AI agents to the content you want them to prioritize. In contrast, robots.txt tells search engine crawlers which pages on your site they have and don’t have access to.
In llms.txt, you can tell the AI which content you want it to view while helping it avoid menus and ads. Unlike robots.txt, however, it’s not yet the standard. There is no guarantee that LLMs that visit your website will follow its rules (not that there is with the other file, but it’s just the “bad crawlers” that ignore it).
So, at this point in time, having an llms.txt file won’t harm your website, but you shouldn’t rely solely on it to improve AI visibility. Many of the major AI models seem to respect it (ChatGPT, Claude, Gemini, etc.), but there is still no solid guarantee.
If you would like to create one for your website, you can write one yourself in Markdown.
Start with an H1 and use H2s for organization.
Add a blockquote under the H1 with a description of your content.
List your most important URLs, each with a concise description. Remember that this is not a sitemap, so only focus on what you want the LLMs to see first.
Upload it to your website’s root directory (public_html, for example).
You can also find guidelines online on how to create a llms.txt file, or just ask your favorite chatbot to generate one for you. Just remember, it’s important to give it more context than just “Make me an llms.txt file.” Watch Nathan’s stream on how to properly prompt your bot if you’d like to learn more.
Track what you can, even if it’s imperfect
As of the writing of this blog post, there isn’t a way to get dedicated “LLM traffic” reports. Yet. It must be right on the horizon, though.
Nonetheless, you can still glean information from the tracking tools we already have. For example, Google Analytics 4 (GA4) can show you the source of traffic. So, if you see chatgpt.com, perplexity.ai, etc., then it’s likely that the traffic is coming from an LLM.
Additionally, Ahrefs (Daphne’s preferred SEO resource) can help you learn about your site’s visibility and mentions across the internet, some of which may come from AI.
Finally, spikes in direct traffic to deep pages, rather than to your homepage, can often indicate increased AI visibility. Since chatbots provide links to users, they bypass your homepage and land directly on the page the AI recommended.
And while none of these methods guarantee that what you see is AI traffic, it’s still good feedback to have on your content.
The bottom line? Play the long game
AI search is evolving before our very eyes, but its foundation hasn’t changed as much as it may seem.
Google and LLMs are not pulling the rug under you, asking you to relearn years of SEO practices. Instead, focus on what already works: genuinely useful content that’s structured clearly.
Experimenting with other ways to improve AI visibility is also highly encouraged, but we caution against overdoing it. The AI search landscape is changing so quickly, that pouring immense time and effort (and money) into something that may change tomorrow isn’t advisable.
It’s simple: if your content is worth reading, it’s worth being surfaced by AI.
FAQ
Does publishing more content faster help with LLM visibility?
Publishing more content quickly tends to work against you, both for traditional SEO and LLM visibility. LLMs, like Google, favor content that demonstrates depth and genuine expertise on a subject. A high volume of thin, hastily produced posts dilutes the authority of a site rather than building it.
A smaller library of well-developed, frequently updated pieces that thoroughly address real questions will consistently outperform a large archive of surface-level content in how both AI systems and search engines surface your site.
Does the age of a domain or website affect how likely it is to be cited by LLMs?
Indirectly, yes, but age itself isn't the mechanism. Older domains tend to have accumulated more inbound links, more mentions across third-party platforms, and a longer track record of publishing useful content, all of which contribute to the authority signals that LLMs draw on when deciding which sources to reference. A brand-new site with exceptional, well-structured content can still get picked up relatively quickly, particularly if it earns mentions on established platforms or fills a gap in available information on a topic.
Should I optimize my content differently for different LLMs?
Not meaningfully, at least for now. The fundamental inputs that make content useful to an LLM (clarity, structure, accuracy, topic depth, and genuine helpfulness) are shared across the major systems. Google's AI Overview draws from Googlebot's indexing, while third-party LLMs have their own training data and real-time retrieval approaches, but none of them have published a separate content specification that differs significantly from well-established SEO and content quality principles.
Can having a Wikipedia page or being cited on Wikipedia improve how often LLMs mention my brand or site?
Yes, and this is one of the more concrete authority signals worth being aware of. Wikipedia is a foundational training source for most large language models, and brands or entities that have established Wikipedia entries tend to appear more reliably and accurately in LLM responses. The same principle extends to other highly authoritative reference sources: industry databases, academic citations, well-regarded journalism, and major directory listings.