AI Overviews, zero-clicks & SEO: how to stay visible in the LLM era
9 minutes
.
October 10, 2025
Since Google I/O 2025, the SERP has undergone a radical transformation. Out go the “10 blue links”, in come AI Overviews, AI Mode and conversational engines like Gemini 2.5, ChatGPT Search or Perplexity.
These new interfaces no longer just link to pages. They compose answers from multiple sources.
What the numbers already show:
- -34% CTR for position 1 since AI Overviews launched, according to Ahrefs (April 2025). Source: Ahrefs, AI Overviews Reduce Clicks
- 4 searches out of 5 end without a click (Bain & Company, May 2025). Source: Bain, Goodbye Clicks
- BrightEdge even reports a +50% increase in impressions, with no matching increase in traffic
Towards a new logic of visibility
The challenge is no longer to be clicked, but to be cited. In this zero-click era, your content pages are no longer destinations. They become building blocks for answers that LLMs use to generate summaries.
SEO is not going away. It is moving to new ground. To exist tomorrow, you need to understand today how to feed AI engines properly.
During the Google I/O 2025 keynote, Sundar Pichai made it clear: the objective is no longer to rank pages, but to compose answers. AI Overviews, AI Mode, and Gemini 2.5 embody this shift to a conversational, AI-driven SERP.
To fully understand this shift, you can (re)watch the official keynote below:
Becoming an LLM-friendly source: the new criteria
To appear in AI Overviews or conversational engines like Perplexity or ChatGPT Search, backlinks and keyword density are no longer what matters most. What counts is clarity, structure and informational value.
Content AI can read... the way a senior consultant would
Language models do not just crawl your pages like a basic crawler. They read, analyze and prioritize content blocks the way a human expert would.
They extract:
- The direct answers to specific questions,
- The useful definitions,
- The sourced and contextualized figures,
- The explicit and concrete methodologies.
Vague, unstructured or overly promotional content will be neither reused nor cited.
What is becoming strategic
The content that surfaces in AI results shares the same fundamentals:
- A logical, hierarchical structure (coherent Hn tags, standalone sections)
- Short, clear paragraphs that can be extracted on their own
- A demonstration of real expertise (examples, figures, reliable sources)
- Original value that AI engines can't come up with on their own
Each section should be able to answer a specific intent. Otherwise, it will be ignored.
SEO vs GEO: two complementary, not competing approaches
In a zero-click context, it would be tempting to abandon traditional SEO in favor of GEO (Generative Engine Optimization). That would be a mistake.
The two approaches serve different, complementary objectives:
- Traditional SEO keeps generating organic traffic through indexed pages and funnel queries.
- GEO, on the other hand, optimizes your visibility in AI-generated answers, where users no longer click... they read.
To go deeper into the SEO vs GEO distinction, keep in mind that these two levers are not mutually exclusive: they reinforce each other in an overall visibility strategy.
An effective SEO strategy in 2025 therefore combines:
- well-ranked pages (structure, internal linking, semantics),
- content designed to be excerpted, quoted and summarized.
Disappearing from AI engines means losing ground. But no longer being clickable in classic SERPs means giving up conversions.
How do you become a source for generative AIs?
In the age of conversational engines, your content is no longer just indexed: it is interpreted, broken down and partially integrated into AI answers. To maximize your chances of being cited, you should structure each page like a knowledge base.
| Recommended practice | AI objective | Why it works |
|---|---|---|
| Clear, hierarchical headings (H2, H3…) | Identify the structure and angles | Makes fan-out easier (splitting into sub-questions) |
| Self-contained, sourced paragraphs | Extract a citation easily | ↑ relevance, ↑ perceived reliability |
| FAQ formats (schema.org / clear Q&A) | Direct understanding of the question | A format AI Overviews use natively |
| Figures with context | Detect usable stats | AI engines favor figures with context |
| Links to related content | Understanding of the topical scope | Strengthens your topical authority |
By applying these principles, you are no longer just writing for Google: you are training AI engines to cite you.
More demanding, more specific, more conversational users
AI Overviews and conversational engines are radically changing the way users interact with search. People no longer type “SEO tool” but “Which SEO tool should I choose for a B2B corporate website with fewer than 10 pages?”
That is the whole logic of query fan-out: Google or ChatGPT break the question down into sub-questions to compose a more relevant answer. Your content should therefore anticipate these specific intents.
This evolution is reflected in three major changes:
- Longer, more contextual queries
Generic searches are clearly declining. Users want advice tailored to their situation, and AI engines favor content that answers it precisely. - Multi-step search
Users ask a first question, get a partial answer, then rephrase or dig deeper. This requires strong semantic linking between your pages and a real content architecture strategy. - A need for immediate authority
Thin content is quickly discarded. What matters now is the ability to reassure AI engines that what you say is reliable: cited sources, figures, concrete FAQs, etc.
Measuring what really matters: from position to authority signal
In traditional SEO, the key metrics were organic traffic, keyword rankings and click-through rate. But in a world dominated by AI Overviews and conversational answers, these benchmarks are losing relevance.
What is disappearing
- CTR & SEO position: a site can stay in position 1 yet be invisible, buried under an AI Overview.
- Number of visitors: if AI summarizes your content without generating a click, your traffic drops... even though you are “present”.
| Post-AI KPI | Why it matters | How to measure it |
|---|---|---|
| AI citation rate | Getting cited by ChatGPT, Gemini or Perplexity | Manual tests on key queries + forum/UX monitoring |
| Residual engagement rate | The few clicks left are highly qualified | Time on page / page views / scroll depth |
| Topical authority | The more a site is cited, the more it becomes a "trusted source" | Mentions in expert content, editorial backlinks |
This shift requires SEO teams to move beyond click-based tracking and adopt a presence-based approach. In other words: am I recognized as a trustworthy source, even without a click?
Conclusion: staying visible means becoming useful
SEO is not going away. It is changing.
In a world where AI generates the answers, your role is to become the raw material.
To do this, it is no longer enough to be well positioned. You have to be structured, quoted and reliable.
Your content must answer complex intents precisely, and your site must guide AI engines as well as humans.
That is why we help businesses rethink their SEO strategy for this new era.
From performance audits to Webflow redesigns and GEO optimization, our approach has one simple goal: make your site an authoritative source that AI naturally chooses.
To find out more about our expertise: Contact us here →
Quick checklist: are you ready for SEO in the LLM era?
| Key criterion | Why does it matter? | What to do in Webflow |
|---|---|---|
| Hierarchical heading structure | Helps LLMs understand the outline of your page | A single H1, logical H2/H3, explicit headings |
| Short, clear paragraphs | Supports automatic extraction and citation | 3–4 lines max, 1 idea per paragraph |
| Self-contained, precise sections | Lets AI isolate useful blocks | Add FAQs, definitions and concrete examples |
| Figures, examples and reliable sources | Gives your content credibility | Cite your sources (Ahrefs, studies, field experience) |
| Semantic tags & structured data | Makes AI indexing and rich snippets easier | FAQ, Article, How-To JSON-LD in the <head> |
| Well-planned internal linking | Guides AI through your site and strengthens topic clusters | 2 to 3 contextual internal links per page |
| Well-structured multilingual version | Avoids cross-market cannibalization and strengthens local visibility | Webflow Localization with hreflang + subdirectories |
FAQ: answering common questions... and AI
How do I know if AI engines are using my content?
Test your topics on ChatGPT, Perplexity, or Gemini: if your key phrases or ideas appear in the answers, that's a good sign. You can also set up AI mention trackers or cross-check queries in Search Console.
Do backlinks still count in an AI context?
Yes, but their role is changing. AI uses backlinks to assess the reliability of a source, not just its raw authority. Five quality backlinks (media, experts) beat 100 low-value links.
Should I continue to track my SEO positions?
Yes, but add other indicators: citation frequency in AI Overviews, paragraph extraction rate, time on page, and performance of your passages included in AI answers.
How do I structure my content so AI engines pick it up?
- Open with a clear answer
- Write each subsection as a standalone answer
- Use clear tags (H2/H3), an educational tone, and specific examples
Is Webflow suited to this new kind of SEO?
Yes. Webflow gives you fine control over HTML, excellent loading speed and precise semantic markup. It is therefore fully compatible with AI requirements, provided you use it well: no misconfigured CMS, no generic content.
