Quick summary: Most B2B teams are working on AEO content while ignoring the technical layer that determines whether any of it gets read. Schema, site speed, content structure, and llms.txt are not optional extras, they're the foundation. Here's what Brittany Trafis, CEO of Soarion Digital, found when she analyzed the AI search presence of B2B companies across three CMO Huddles Strategy Labs.
There is a rumor going around the AEO space that the fix is more content. Write more. Publish more. Cover more topics. If you just pump enough content out, the LLMs will find you.
Brittany Trafis, CEO of Soarion Digital, has heard this in every room she walks into. And she has the data to show why it's wrong.
Trafis runs an AI-native, AI search agency. Before launching Soarion Digital, she and her co-founder ran a traditional digital marketing agency for over a decade, sold it, and rebuilt from scratch with AI at the center. Across three CMO Huddles Strategy Labs, she analyzed the AI search presence of the companies in the room and found the same pattern in every city: Solid SEO foundations, poor AEO technical readiness, and content that the LLMs simply could not get to.
“You can write all the content all day if you want,” Trafis told a room of senior B2B CMOs. “But if the AI can’t get to it, then it doesn’t matter.”
The Content Readiness Problem
Across her three sessions, Trafis scored each room’s websites on content readiness, which she defines not as content quality but as the readiness of that content for AI agents to crawl, retrieve, and use. The room average was 45 out of 100.
That score does not mean the content was bad. It means the LLMs kept getting stuck. Too many heavy pages. Too little structure. No schema telling the agent what the page was about, no direct answer blocks to pull from, and no FAQ sections to match against incoming prompts.
The companies in the room had spent years building for two audiences: Humans and Google. They had not yet built for the third.
“We built our websites for humans, which was important because that’s who was engaging with them. But now we have a new audience too. We have humans and we have machines that we really have to start writing for and building for.”— Brittany Trafis, CEO, Soarion Digital
The Fix That Moves the Needle Fastest: Schema Plus FAQs
According to BrightEdge research that Trafis cites across all three sessions, adding schema and FAQs to your top pages produces a 44 percent increase in citation rate. That is the single highest-leverage technical action available to most B2B marketing teams right now.
Here is why it works. When an LLM receives a prompt, it goes into retrieval mode in milliseconds. It is scanning for content it can pull quickly and deliver cleanly. Schema sits in the back end of your page and tells the agent exactly what it is looking at, without having to parse through all the surrounding content. FAQ sections give it pre-formatted question-and-answer pairs that map directly to the kinds of prompts buyers are typing.
The combination means the agent gets what it needs fast, cites your page, and moves on.
44% increase in citation rate From adding schema and FAQs to your top pages, according to BrightEdge research cited by Trafis across multiple Strategy Labs.
A few things to get right:
- Use the correct schema type for each page. Blog posts need article schema. FAQ sections need FAQ schema. Using the wrong type reduces the benefit.
- FAQs should not just regurgitate what is already on the page. The LLM already knows what's on the page. Write questions that buyers are actually prompting, mapped to the content on that specific page.
- Four to five strong FAQs per page is the right range. More than that creates noise. Fewer than that leaves prompts uncovered.
- Use an LLM to generate FAQ drafts, but review every one with a human. Otherwise, you end up with the same generic questions as every other site in your category.
AI Readiness and LCP Speed
LCP Speed is the technical issue Trafis finds most consistently overlooked, and the one with the most dramatic gap between where teams think they are and where they actually are.
LCP stands for Largest Contentful Paint. It measures how long it takes for the main content of a page to load. The threshold that matters for AI search is 2.5 seconds. Across three CMO Huddles Strategy Labs, room averages ranged from 8.5 to 12.5 seconds.
That gap is not a small problem. It is the reason LLMs are skipping entire websites.
2.5 sec LCP threshold for AI search. The target load time for LLMs to engage with your content. Room averages across Strategy Labs ranged from 8.5 to 12.5 seconds.
Trafis uses a specific analogy to explain what is happening: Imagine you are a researcher with 100 books to get through. You pick up the first one and the cover weighs 100 pounds. Are you going to keep fighting it? Or are you going to put it down and move to the next book? That is exactly what an LLM does when your page takes too long to load. It moves on to your competitor.
What is causing the weight? The beautiful hero banners your design team spent weeks on. The background videos. The heavy images that look stunning on desktop and take forever to load on the back end. They are built for humans who will wait. LLMs will not.
⚠️ If you are on a CMS like WordPress, HubSpot, or Webflow, your web team can check your LCP score in Google PageSpeed Insights in minutes. Run it now before your next content sprint. If you are over 2.5 seconds, fixing the load time will do more for your AEO performance than any amount of new content.
One additional flag from Trafis: If your CEO has ever arrived on a Monday morning with a website they built over the weekend in a no-code tool, check the load time before it goes anywhere near your domain. Those sites are notoriously slow. Speed is not optional; it's a gating factor.
The llms.txt File: The Fastest Technical Fix
Less than 20 percent of B2B websites currently have schema. Even fewer have an llms.txt file, and most of the CMOs in Trafis’s AEO sessions had never heard of one.
The llms.txt file is a plain text file that lives on your website and gives AI agents a direct briefing on who you are, what you do, and what you want them to know. Think of it as the LLM equivalent of a meta description, but written for agents rather than for humans or Google.
You can link to it from your site footer. An agent crawling your site sees it, reads it, and has a clean summary of your company before it even gets into your pages. It takes one session to write and one hour to implement.
Critical instructions from Trafis:
- Do not write it like marketing copy.
- Don't say things like “best platform” or “leading solution.” LLMs are smart enough to recognize sales language and will ignore it.
- Write it the way a neutral analyst would describe your company. What problems do you solve? Who do you solve them for? What makes you different?
- Keep it factual, specific, and short.
How to Structure Content the LLM Can Actually Use
Beyond schema and speed, the way content is written and structured on the page matters significantly. Trafis describes a specific format she calls the trifecta: Writing for humans, for Google, and for AI simultaneously. The structure is:
- H1 that answers or directly addresses the prompt you want to show up for
- A 60-word direct answer block immediately after the H1. Not 80 words. Not 100. Sixty is the number Trafis has tested across client work, and it's where LLMs reliably pull the answer and cite the source
- H2 subheadings with supporting content in bullets and tables rather than dense paragraphs
- FAQs at the bottom with question-format entries that map to related prompts
The 60-word block is the piece most teams miss. When an LLM gets a prompt, it fans out across sources and looks for a clean, direct answer it can lift and deliver. If it finds one near the top of your page, it takes it, cites you, and stops reading. If it does not find one, it keeps scanning. You want to be the source that stops the scan.
One practical note on refreshing old content: Adding a 60-word direct answer block, updating the headline, and adding FAQs to an existing page counts as a meaningful refresh in the eyes of LLMs. You do not need to rewrite the whole page. This matters because Trafis also reports that approximately 50 percent of content being served in LLMs is less than 13 weeks old. Your two-year-old SEO star is likely being deprioritized. Refreshing it with the right structure can bring it back.
The robot.txt File and What It Actually Does
Most B2B websites already have a robot.txt file in place, which was encouraging news across Trafis’s AEO session. What it does is simple: It tells crawlers whether they are allowed to access your site. A year and a half ago, many companies were blocking AI crawlers out of concern about data use, but the posture has shifted. You actually want agents to come in.
The robot.txt file is not the same as the llms.txt file. Robot.txt says yes, you can crawl. The llms.txt file says here is what I want you to know when you do. Both matter. Both are quick to implement. Most teams only have one.
Where to Start: The Prioritized Checklist
Trafis is consistent across all three sessions on sequencing. Do not try to fix everything at once. Pick your top 25 pages, the ones most important to the clusters you want to own, and work through this list in order:
- Check your LCP speed in Google PageSpeed Insights. If you are above 2.5 seconds, brief your web team on compressing hero images and reducing page weight.
- Add schema to your top 25 pages. Use the correct type for each: Article schema for blog posts, FAQ schema for FAQ sections, organization schema for your homepage.
- Add four to five FAQs to the bottom of each of those pages. Write them in buyer language, mapped to prompts, not to what is already on the page.
- Add a 60-word direct answer block at the top of each page, immediately after the H1.
- Create an llms.txt file and link to it from your footer. Write it in plain, factual language.
- Check that your robot.txt file is not blocking AI crawlers.
- Refresh your highest-performing legacy pages with the above structure and update the publish date.
The timeline for results is faster than most teams expect. Trafis reports seeing citation rate movement in as little as one week after implementation, with an average of two to three weeks across her client base.
Brittany Trafis is CEO and co-founder of Soarion Digital, an AI-native AI search agency. She facilitated three CMO Huddles AEO Strategy Labs sessions in 2026.
Want more?
- For the strategic case for AEO and what it means for B2B marketing, read: The 4 Pillars of AEO: A Framework to Win AI Search.
- For the tactical guide to topic clusters, buyer language, and content strategy, read: B2B AEO Strategy: The Tactical Guide to Getting Found in AI Search.
- CMO Huddles brings together senior B2B marketing leaders for candid, peer-to-peer conversations on the challenges that matter most. Starter membership is free.
Frequently Asked Questions About AEO Technical Readiness
AEO technical readiness refers to how easily AI agents can crawl, retrieve, and use the content on your website. A site can have excellent content and still score poorly on technical readiness if it loads too slowly, lacks schema markup, has no direct answer blocks, or stores key content in PDFs or behind forms. Technical readiness is the layer that determines whether your content strategy reaches LLMs at all.
LCP stands for Largest Contentful Paint. It measures how long it takes for the main content of a page to load on the back end. The threshold that matters for AI search is 2.5 seconds. Above that, LLMs are likely to skip the page and move to a competitor. Across three CMO Huddles Strategy Labs, room averages ranged from 8.5 to 12.5 seconds, meaning that many B2B websites are significantly over the threshold. The primary culprits are heavy hero images, background videos, and JavaScript-heavy pages.
According to BrightEdge research cited by Brittany Trafis, adding schema and FAQs to your top pages produces a 44 percent increase in citation rate. Trafis reports seeing results in as little as one week after implementation, with an average of two to three weeks. The key is doing both together: Schema tells the agent what the page is about, FAQs give it pre-formatted answers to match against incoming prompts.
An llms.txt file is a plain text file linked from your website footer that gives AI agents a direct briefing on your company. It is similar in concept to a robots.txt file, but written for LLMs rather than crawlers. It should describe who you are, what problems you solve, who you solve them for, and what makes you different, in plain factual language rather than marketing copy. LLMs recognize sales language and ignore it. A well-written llms.txt file can be created in one session and implemented in under an hour.