A client called me last month, half-annoyed and half-curious. He’d asked ChatGPT to recommend suppliers in his industry. It named four companies. His was not one of them, and two of the four were smaller than him with worse websites.
He wanted to know what he did wrong. The honest answer is that he didn’t do anything wrong. He just spent fifteen years optimizing for a version of search that is quietly being replaced.
The click is disappearing
Here’s the number that should get your attention. Between January and April 2026, 68% of U.S. Google searches ended without a single click on a result. That’s up from 60.45% in 2024, according to SparkToro’s analysis of Similarweb clickstream data.
Two-thirds of searches now end on the results page. People read the AI summary, get their answer, and move on. Your beautifully written service page never enters the picture.
This isn’t only a Google problem. People are asking ChatGPT, Perplexity, Gemini, and Claude for recommendations the way they used to type queries into a search box. When those tools answer, they name a handful of businesses. If you’re not in that handful, you’re not in the conversation at all. There’s no page two to fall back on. We wrote more about this shift in where AI-run websites are heading.
That’s the shift. Search used to rank you. AI answers either mention you or they don’t.
Ranking well and getting cited are different problems
The instinct is to assume this sorts itself out. Rank on page one and the AI will pick you up. It doesn’t work that cleanly.
Research from Conductor on how different engines choose sources found that each one has what they call a persistent editorial identity, a default source type it reaches for. ChatGPT Search leans heavily on Wikipedia when someone asks for education, comparisons, recommendations, or purchase advice. Perplexity and Google’s AI Overviews reach for YouTube across most intent types. Google’s AI Mode routes purchase questions toward Shopping and Maps.
Three different Google products behave three different ways despite sharing infrastructure. So the old idea of one SEO checklist covering everything is over. You now need to think about where each engine actually looks.
For a small business in the Philippines, that usually means your Google Business Profile and any third-party page that mentions you by name carry more weight than another 2,000-word blog post nobody links to.
What actually moves the needle
I’ll go through what we’ve seen work on client sites, and I’ll be specific about what we’ve seen fail.
Answer the question in the first hundred words
AI systems extract passages, not pages. If your service page opens with a paragraph about your commitment to excellence, there’s nothing to pull. Lead with the direct answer. What you do, who it’s for, what it costs, where you are. Then elaborate below.
Give every page a clean structure
Real headings that describe real sections. Short paragraphs. Tables for comparisons, lists for steps, a FAQ block for the questions clients actually ask on calls. Extractable content gets cited. Wall-of-text pages don’t.
Add proper schema markup
Organization, LocalBusiness, FAQPage, Product, Article. This is the part most WordPress sites get half-right because a plugin generated something generic years ago and nobody checked it since. Wrong or incomplete schema is common and it’s cheap to fix, and it is one of the first things a WordPress website audit should catch.
Get mentioned somewhere other than your own site
This is the uncomfortable one. AI models weight what other sources say about you, so a mention in an industry directory, a local news piece, a Reddit thread, or a genuine customer review often does more than a new page on your own domain. You can’t optimize your way into a citation nobody else corroborates.
Check that you’re not blocking the crawlers
I’ve audited sites where the owner had no idea their security plugin or CDN was refusing GPTBot and PerplexityBot. Look at your server logs. If those bots aren’t showing up, nothing else you do matters.
Keep your basic facts consistent and current
Same business name, address, and phone number everywhere. Dates on your posts. A real author with a real bio. AI systems cross-check, and contradictions make you a risky source to cite.
What doesn’t work, despite the hype
You’ve probably seen posts telling you to add an llms.txt file to your site. It’s presented as the AI-era robots.txt, the thing that finally gets the models to read you properly.
Search Engine Land tracked ten sites for 180 days, 90 before adding the file and 90 after. Eight saw no measurable change. One dropped 19.7%. The two that grew turned out to have other explanations: one got Bloomberg coverage and restructured its product pages in the same window, the other had published 27 downloadable templates three weeks earlier.
“None of the AI services have said they’re using llms.txt, and you can tell when you look at your server logs that they don’t even check for it.” — John Mueller, Google
Adoption backs this up. As of June 2026, 87 of the top 1,000 websites publish one. It costs nothing to add, so add it if you want. Just don’t treat it as your AI visibility strategy, and don’t pay anyone a retainer for it.
How to check where you stand, today
You don’t need a tool or a consultant to run this first check. Fifteen minutes will tell you most of what you need to know.
Open ChatGPT, Perplexity, and Google AI Mode. Ask each one the question a customer would actually ask. Not “tell me about [your company],” because that’s a lookup, not a recommendation. Ask “who are the best web design companies near Clark, Pampanga” or whatever the equivalent is in your industry.
Write down who gets named. If it’s never you, that’s your baseline. If your competitors show up consistently, go look at what those pages have that yours don’t. Usually it’s clearer structure, or reviews and mentions on sites they don’t own.
Then pull your server logs and search for GPTBot, PerplexityBot, ClaudeBot, and Google-Extended. If they’re absent or getting 403s, that’s your first fix and it’s usually a five-minute change.
The part nobody wants to hear
There’s no trick here. The businesses getting cited by AI tend to be the ones with accurate information, clear pages, real reviews, and a presence beyond their own domain. That’s not new advice dressed up in new letters. It’s the same fundamentals, except the penalty for ignoring them got a lot steeper, because there’s no page two to rank on anymore.
What is new is that most sites were never built with extraction in mind. They were built to look good and to rank. Those aren’t the same job, and the gap between them is where your competitors are winning right now.
At Cognitech Digital Studio we’ve started running AI visibility as part of every audit, including our free WordPress audit: checking what the major engines say about a business, fixing the schema and structure, unblocking the crawlers, and finding the off-site gaps. Sometimes the finding is that everything is fine and the client just needs more reviews. Sometimes we discover a firewall rule has been quietly turning away every AI crawler for eight months.
Either way, you’d want to know.
If you’re not sure where your site stands, run the fifteen-minute check above. If you’d rather have someone look at it properly, book a free consultation and we’ll go through it with you.