SEO & AEO Engine — Visible in AI Search

This isn't a ranking problem you can fix. It's a redistribution of labor: AI systems now do the finding.

When an AI overview appears for a search query, click-through rates on the classic result list collapse. Only 8% of users still clicked a classic result when an AI overview was shown, versus 15% without one — and only 1% clicked through to any of the sources cited within the AI answer at all.

According to Ahrefs (February 2026, analysis of 300,000 keywords), the click-through rate for position 1 fell from 7.3% to 1.6% once an AI Overview appeared — a 58% decline after adjusting for general trends.

The Pew Research Center reached a similar conclusion in its own study (July 2025, roughly 69,000 search queries): only 8% of users still clicked a classic result when an AI overview was shown, versus 15% without one — and only 1% clicked through to any of the sources cited within the AI answer at all.

Similarweb measures zero-click searches rising from 56% to 69% between May 2024 and May 2025; SparkToro puts the US figure at 68% in early 2026, up 7.6 percentage points from 2024.

What follow-through actually means

Anyone who still clicks through to a company website today has usually already had the easy question answered — the AI took care of that. What's left is the remainder: the question too specific, too individual, or too case-dependent to answer generically. That visitor isn't looking for more information. They're testing whether the conversation continues, or whether it resets back to a page they now have to search through themselves.

AEO - The invisible knowledge layer

The data you already provide us for your knowledge base gets put to double use: once so leo.page can advise your visitors — and once to automatically generate an invisible, cleanly marked-up HTML layer built specifically for AI search systems to read. These pages aren't part of your visible website and aren't linked anywhere — but they're discoverable for search and AI crawlers via your sitemap.

When you're already found, follow-through decides the rest

There's one exception to falling click-through rates: searches that already contain the company name. Here, AI overviews show the opposite pattern — up to 18% higher click-through rates when the brand is named in the answer. Being found, in that case, isn't the problem — it's already solved. What decides the outcome from there is exactly what this page is about: whether the follow-through holds up once they arrive.

Book demo

Try leo.page

Why is my website losing traffic even though I haven't done anything wrong?

It's rarely the site's own fault. Google and other search systems increasingly answer queries directly in an AI overview, without requiring a click — according to SparkToro, that now affects roughly 68% of all US searches. Being found has quietly stopped being something a website wins or loses on its own.

Does this also apply to the European B2B market?

The solid numbers come mostly from the US and from large publishers. For European B2B niche markets, the effect is plausible but not yet independently measured — the honest answer is: probably present, but slower and weaker than the US data suggests.

Do I have to react to this, or can I react to this?

You can. Being found is already shifting to AI systems whether you act or not — that part isn't a choice. What is a choice is how deliberately you handle the follow-through once someone shows up anyway, and on what timeline. Nothing breaks if you wait a quarter. But this isn't a one-off spike either — the underlying numbers (fewer clicks, more zero-click searches) have been moving in the same direction for two years running, so waiting has a cost that compounds quietly rather than one that hits all at once. Treat it as a shift worth planning for, not an emergency.

Is this why chatbots are booming?

Partly. The same shift explains both trends: people expect a conversation to continue a conversation, not to hand them back a page of text. But most chatbots only address the surface of that expectation — they repeat information faster, without advising. That's not follow-through, it's just faster information. leo.page is built specifically for the follow-through itself: it asks the next question, clarifies the actual case, and reasons through it the way a good salesperson would — which is also why it's built and positioned deliberately not as "a chatbot."