You are paying for SEO, but when you open your analytics the inbound leads are not there. Meanwhile your customers have started asking ChatGPT for recommendations, and your name never comes up. That gap is the whole problem, and it is fixable.
AI search is no longer a thing to watch for later. It is already a measured channel. Google Analytics now reports an "AI Assistant" traffic source right alongside Organic Search and Direct, and your own site analytics will show it too. People are running searches inside ChatGPT, Claude, Grok, and Google's own AI results instead of scrolling a page of blue links. Build your site so those models can read it and trust it, and you become one of the answers they hand back. Skip that work and you stay invisible, no matter how good the service itself is.
Below are the four things you need on your site to start showing up. The first is about substance. The other three are technical, and I will keep them plain enough to hand to whoever manages your website.
Watch: the four things, straight from me, in a couple of minutes.
What we actually saw with one client
We took over a client, Collaborate Real Estate Group, that had been paying for "SEO" for a while. When we looked at the analytics, there was almost no measurable organic traffic and no inbound leads to show for the spend. So we changed the approach. Instead of generic keyword pages, we started publishing real stories from their team members and answering the actual questions their audience was asking.
The numbers moved. Organic search is now their highest-engagement channel, time on site is up, and they have pulled in four inbound leads in the last three weeks. That last number is small, and I want to be honest about that, but it went from zero, which is the part that matters. The most interesting line in the report, though, was the new AI Assistant channel.
Visitors arriving from AI search spent 42 seconds engaged on the page, with a 75% engagement rate and 6.13 events per session.
That was the highest-quality traffic on the entire report: roughly 42 times the time on page of paid social and about five times longer than direct visits. The volume was small (8 sessions), but every one of those people came pre-qualified by the AI before they ever landed. Source: Collaborate Real Estate Group, Google Analytics 4, last 90 days (Mar 31 to Jun 28, 2026).
Eight visits will not change a business on their own. But the behavior behind them is the signal worth paying attention to. When an AI assistant recommends you, the person who clicks through already trusts the answer, so they read, they poke around, and they show up far closer to reaching out. That is the traffic you want more of, and here is how you earn it.
More buyers now start with an AI assistant instead of a page of blue links.
1. Use your real expertise to answer real questions
This is the part most businesses get wrong, and it is the one that matters most. AI models quote answers. So your site needs to answer the questions your customers are actually asking, but it cannot be a brag, and it cannot be a generic AI-written listicle that anyone could have produced. We have been preaching this for years, and AI search has only made it more true.
The standard to write to is E-E-A-T: Experience, Expertise, Authority, and Trustworthiness. In plain terms, that means taking your real stories and your hard-won experience and using them to answer the question in a way a competitor cannot copy. If your blog post could have been typed by anyone with a keyboard and a prompt, it will not stand out, and the AI has no reason to pick you over the thousand other pages that say the same thing. Tell the story. Share what actually happened on a project. Connect to your reader as a person who has done the work. That is the easiest way to become the page the AI decides to cite.
The thing AI cannot fake is your experience. That is exactly why it is the thing worth putting on the page.
This is not just a hunch. HubSpot found that content carrying strong experience and trust signals is about 96% more likely to get cited by an AI than content that reads as generic. Source: HubSpot.
Not sure whether your homepage answers the questions a buyer (or an AI) is actually asking?
Run the free 5-question homepage audit →2. Build an FAQ section, then put a name tag on it
Once you are answering questions, gather them into a real FAQ section on your site. Find out what people are asking on Google and inside the AI search engines, and answer those questions using the E-E-A-T approach from step one.
Here is where it gets a little technical, and I will make it simple. That FAQ section needs a name tag. In the coding world it is called schema, and it is a small piece of invisible code that sits in the background of the page. A visitor never sees it. What it does is point a big red sign at your answers and tell Google and the language models: this right here is a question, and this is its answer. It removes the guesswork, which is exactly what an AI wants.
The good news is you usually do not hand-code this. If you are on Wix or Squarespace, FAQ schema is often handled for you automatically. If you are on WordPress and you used a proper FAQ block or template, the schema is likely already baked in. You just need to confirm it is there. Three schema types do the heavy lifting: FAQPage for your questions and answers, Article for your blog posts, and Organization for your homepage. Put FAQ schema on your most important pages first.
3. Tell Google and the AI who you are
As you build all of this, you also need to tell the search engines and the AI models who is behind the content. Think of it as putting a business card on your homepage and on every piece of long-form writing you publish. On your articles, clearly define who wrote it, and make that a real person with a name, a title, and a position, not a vague "written by the company." A named human with relevant experience adds credibility to the page, and credibility is what makes an AI comfortable repeating your answer to its user.
There is one more small technical piece here: list the updated or modified date on your articles. When someone runs a search, the engines and the AI models favor the most recent, most current content. A clear "last updated" date signals that your information is fresh and worth trusting. In testing, when the author, the organization, and the article are all properly connected, AI citations have gone up by as much as three times. Source: Search Engine Land data, reported via Presence AI (February 2026).
4. Add an llms.txt file: a roadmap built for AI
The fourth thing is the most technical of the group, and it may also be the easiest to do. You add an llms.txt file to your site, hosted at the root, so it lives at yourbrand.com/llms.txt. It is brand new. We know Anthropic publishes one. The others, including OpenAI and Google, have not confirmed they consume it yet, so I want to be straight with you: this is a smart, cheap bet, not a magic switch.
That said, given how fast things are moving, and the fact that Google is now putting an AI chat experience inside Google Business Profile, it is reasonable to expect wider adoption is coming. The file itself is basically a roadmap of your site. It clearly lays out where everything lives, which pages are the parent pages, which are the child pages, and how they connect, so an AI model knows exactly where to go to find the information you want it to share. As technical as that sounds, you can have an AI tool generate the file for you in a few minutes. It is low effort and low cost, and it could be one of the small moves that changes how your information gets found. For context, only about 10% of websites publish one today, so doing it now puts you ahead of most of your market. Source: 20North Marketing.
Free download: The AI Search Readiness Checklist. One page that walks through all four steps, including the exact schema types to confirm and a sample llms.txt structure you can copy.
Get the checklist →What these four things add up to
None of this is about chasing a trend. It is about making your website do real work. When you answer real questions with real expertise, tag those answers so machines can read them, tell the AI who you are, and hand it a map of your site, your website stops being a page you hope people land on. It becomes a place you actively guide people to, which is exactly what turns it into a tool your sales team can use to close more clients.
More people are searching inside ChatGPT instead of Google every month, and Google is folding AI answers into its own results. The businesses that set their sites up now are the ones the AI will be recommending a year from now. The four steps above are how you get on that list.
Here are all four steps in one place. Save this section or hand it straight to whoever manages your site.
The four-step AI search checklist
- 1Answer real questions with real experience
Write to the E-E-A-T standard and use your own client stories to answer what your audience is actually searching for. Skip the generic, AI-written listicle that anyone with a prompt could have produced.
- 2Build an FAQ and tag it with schema
Gather the real questions people ask and answer them in a dedicated FAQ section. Then confirm FAQPage schema is active on your most important pages. Do not assume it is there, check it.
- 3Tell the AI who you are
Name a real human author with a title on every article, add Organization details to your homepage, and show a modified date so your content reads as current and trustworthy.
- 4Add an llms.txt roadmap
Host a plain llms.txt file at the root of your site that lists your important pages, and how the parent and child pages connect, so an AI knows exactly where to look. Low effort, low cost.
