93% of AI Mode Searches End Without a Click: How to Build Pipeline When AI Eats the Traffic in 2026
About 93% of Google AI Mode sessions in Semrush's 2025 US desktop data ended without an external visit. Here is the operator playbook for building pipeline when AI intermediates buyer discovery — backed by arxiv research, Forrester data, and real conversion metrics.
About 93% of Google AI Mode sessions end without a visit to an external website, according to Semrush's clickstream analysis of almost 69 million Google Search sessions (US desktop, May-July 2025). That is not a rounding error. It is a structural shift in how buyers find, evaluate, and choose vendors — and most companies are still running a playbook built for the 7% that clicks. The question is no longer how to win the click. It is how to build pipeline inside the answer itself.
I have been watching this happen for eight years. The companies that will own the next decade of B2B growth are not the ones with the best SEO. They are the ones whose brands are already embedded in the answers AI gives when a buyer asks who to trust.
What the 93% AI Mode Zero-Click Rate Actually Measures
The statistic comes from Semrush's clickstream analysis of AI Mode sessions between May 1 and July 5, 2025, when AI Mode was roughly 1% of Google search sessions. When Google AI Mode generates a synthesized answer, only 6-8% of sessions produce an outbound click. In most sessions the user stays inside Google's interface, reading the answer or asking a follow-up that the AI answers again.
Semrush's separate, earlier study of Google Search measured about 34% zero-click without an AI Overview and about 43% with one. Semrush notes that the figures come from different studies, methodologies and timeframes, and reads the overall trend as clicks declining as Google integrates more AI-driven responses. A paper accepted to ACM SIGIR 2026, from researchers at the New Jersey Institute of Technology, Nanyang Technological University and Indiana University, examined how generative AI search retrieves and presents information differently from traditional search. Its 11,500-query benchmark found AI Overviews on 51.5% of its real-user (ORCAS) queries and substantially different sources across Google results, AI Overviews and Gemini Flash 2.5 (average Jaccard similarity under 0.2). It does not measure clicks. It shows that what appears inside the answer is assembled differently from what the blue links rank.
The traffic pattern is shifting beneath everyone's feet. HUMAN Security's 2026 State of AI Traffic report found that monthly AI-driven traffic grew 187% from January to December 2025, nearly tripling over the calendar year.
The same report found automated traffic growing eight times faster than human traffic, year over year. The machines are arriving faster than people are — and they are deciding whether to cite you, not whether to click.
Why Zero-Click Does Not Mean Zero Influence
Here is where most analysis of this stat goes wrong. The 93% figure does not mean your brand has no opportunity to reach the buyer. It means the opportunity moved upstream — into the answer itself.
Forrester's research confirmed this in January 2026: 94% of B2B buyers now use generative AI or conversational search, and twice as many named it as a more meaningful or important source of information than any other source — including vendor websites, product experts, and sales reps. The buyers are not gone. They are asking AI instead of clicking through your funnel.
And when AI does send traffic, it may be worth more per visit than organic clicks. VentureBeat quoted Wyatt Mayham of Northwest AI Consulting, who said LLM-referred traffic converts at 30 to 40% for his own company. That is one firm's figure, not a market rate or an enterprise benchmark. That is not a cold click. That can be a warm handoff.
Meanwhile, TechCrunch reported that AI traffic to U.S. retailers rose 393% in Q1 2026, and it is boosting revenue. The companies capturing this traffic are the ones AI engines already trust as sources.
The Real Problem: Your Funnel Was Built for Clicks That Are Not Coming
Most B2B marketing stacks are optimized for a click-through journey: ad or organic result → landing page → form fill → nurture sequence → sales call. Every layer depends on the initial click.
For buyers who research in AI Mode, where 92–94% of sampled sessions never produced that click, that architecture breaks at its first step. It does not matter how good your landing page is if the buyer never arrives. It does not matter how sharp your nurture sequence is if the buyer formed their shortlist inside ChatGPT before they ever saw your brand name.
Forrester identified this blind spot directly. Their research on zero-click buyer data hiding in bot traffic found that business buyers are researching brands long before they click — and increasingly, they are not clicking at all. The buying decision is forming inside AI-generated answers, and most companies have no visibility into it.
A separate SIGIR 2026 log analysis of 14.44 million agentic search requests, from an open-source search API used by external agentic clients, found that agents research in multi-step sessions and that their later queries are often traceable to evidence they retrieved earlier (on average 54% of newly introduced query terms). It does not measure what buyers click. It does show that when an agent does the researching, the evidence it retrieves shapes what it asks next, so being in that evidence matters at every step.
How AI Decides What to Cite in Zero-Click Answers
Understanding what AI engines cite is now a pipeline question, not a marketing curiosity.
A 2026 study measuring Google AI Overviews, accepted to the ACM Internet Measurement Conference, describes AI Overviews as synthesizing and delivering a single answer, which gives Google editorial control over what users read. Across 55,393 trending queries it found AI Overviews on 13.7% of queries overall (64.7% of question-form queries), cited domains more credible than the co-displayed first-page results, and nearly 30% of cited domains absent from those first-page results altogether, which points to a source selection mechanism distinct from Google's ranking.
Here is our working hierarchy of what AI engines select, a synthesis of our own editorial judgment rather than a finding of that study:
| Source Type | AI Citation Likelihood | Why |
|---|---|---|
| Earned media in trusted publications | Highest | Third-party validation + editorial quality signals |
| Primary research with named methodology | High | Specific, verifiable, structurally extractable |
| Brand-owned content with entity clarity | Medium | Useful if structured, but no third-party trust signal |
| Generic blog content and SEO filler | Low | AI engines deprioritize undifferentiated owned content |
| Paid placements and advertorials | Lowest | AI engines distinguish commercial intent from editorial authority |
The pattern is consistent across ChatGPT, Perplexity, Gemini, and Google AI Mode. The brands getting cited are the ones with earned media placements in publications that AI engines already trust as authoritative sources.
The 5-Move Playbook for Building Pipeline Without Clicks
If your current pipeline depends on organic click-through, here is the operator-level shift required to build pipeline in a zero-click environment.
1. Audit Your AI Visibility Before Optimizing Anything
You cannot fix what you cannot measure. Before changing your content strategy, ask every major AI engine — ChatGPT, Perplexity, Gemini, Google AI Mode, Claude — the buyer queries that matter most to your category. Document whether your brand appears, whether it is cited, and what sources are being cited instead.
Most companies discover they are invisible in AI-driven discovery even when they rank well in traditional search. These are different systems with different source selection criteria.
2. Build Earned Authority Through Trusted Publications
Earned media placements in publications that AI engines trust — Forbes, TechCrunch, Harvard Business Review, industry-specific journals — are the single highest-leverage input for AI citation. This is not about vanity press hits. It is about getting your claims, your methodology, and your brand name embedded in the sources that AI systems pull from when answering buyer queries.
Search Engine Land's analysis confirms this directly: PR is becoming more essential for AI search visibility because the same publications that shaped human brand perception for decades are now the publications AI engines treat as authoritative.
3. Structure Your Content for Extraction, Not Just Reading
AI engines do not read your content the way a human does. They extract specific claims, data points, and definitions. If your content is not structured for extraction — clear headings, direct answers, cited statistics, comparison tables — AI engines will skip your page and cite a competitor whose content is easier to parse.
Peer-reviewed research on AI citation behavior shows that structured, citation-ready content with specific claims and named sources gets extracted at higher rates than narrative-only content. The gap is not small.
4. Track Bot Traffic as a Pipeline Signal
Forrester's research makes this operational: the zero-click buyer data hiding in your bot traffic is your new top-of-funnel signal. ChatGPT-User, PerplexityBot, OAI-SearchBot, and ClaudeBot are visiting your pages to evaluate whether to cite you. That bot traffic is the leading indicator of whether your brand will appear in the AI answers your buyers see.
Companies that track and optimize for AI bot retrieval patterns will see the pipeline impact before companies that only track human clicks.
5. Optimize for Citation, Not Just Ranking
Traditional SEO measures success by ranking position and click-through rate. In an environment where most AI Mode sessions never click out, the metric that matters is citation architecture — whether AI engines cite your brand, your methodology, and your claims when answering the queries your buyers ask.
The optimization target is different from anything SEO taught you. Forget keyword density and backlink profiles. Focus on entity clarity, source credibility signals, and structured extractability across every surface where AI engines look for answers.
The Click-Through Model vs. the Citation Model
| Dimension | Click-Through Model (Legacy) | Citation Model (2026) |
|---|---|---|
| Success metric | Clicks, CTR, rankings | Citations, recommendations, share of AI answers |
| Primary channel | Organic search results | AI-generated answers across ChatGPT, Perplexity, Gemini, Google AI Mode |
| Traffic source | Direct clicks from SERPs | VentureBeat quoted Wyatt Mayham of Northwest AI Consulting, who said LLM-referred traffic converts at 30 to 40% for his own company. That is one firm's figure, not a market rate or an enterprise benchmark. |
| Trust signal | Backlinks, domain authority | Earned media, editorial placement, third-party validation |
| Content requirement | SEO-optimized pages | Structured, extractable, citation-ready content |
| Pipeline mechanism | Click → Landing page → Form fill | AI citation → Warm referral → Buyer who has already seen the AI answer |
| Measurement blind spot | Assumes all discovery starts with a click | Tracks bot traffic, citation presence, and AI answer share |
What This Means for B2B Pipeline in 2026
Forrester put it plainly: GenAI is rebuilding search, and Google is still winning — Q1 2026 search revenue was up 19% year-over-year even as AI eats the click. The economics of search are shifting from click-based to citation-based, and the companies winning are the ones that already understood the mechanism.
Earned media placements in trusted publications are not a PR tactic. They are a pipeline input. When a buyer asks ChatGPT or Perplexity who leads your category, the answer is downstream of your editorial presence — not your ad budget.
This is what Machine Relations defines as the operating discipline for the AI era. Not SEO rebranded. It is the recognition that AI-mediated discovery requires a different source architecture entirely: earned authority as the foundation, entity clarity as the structure, citation across AI answer surfaces as the measurable outcome. The publications have not changed. The mechanism has not changed. The reader changed. The reader is now a machine.
The brands that will own pipeline in 2026 are not fighting the 93% zero-click rate. They are building inside it. They are the answer the AI gives, not the link the AI sometimes includes.
Run a free AI visibility audit to see exactly where your brand appears — and where it is invisible — in AI search.
Frequently Asked Questions
What does the 93% AI Mode zero-click rate mean for B2B companies?
It means that in Semrush's US desktop clickstream sample (May to July 2025), about 93 out of 100 Google AI Mode sessions ended without a visit to an external site. For B2B companies, this means buyers are forming vendor shortlists inside AI answers before ever visiting a website. The opportunity is not in the 7% that clicks — it is in becoming the source the AI cites when it answers the query. Companies that track AI visibility and optimize for citation are capturing pipeline from the 93%.
How does LLM-referred traffic compare to organic search traffic?
VentureBeat quoted Wyatt Mayham of Northwest AI Consulting, who said LLM-referred traffic converts at 30 to 40% for his own company. That is one firm's figure, not a market rate or an enterprise benchmark. That is one firm's figure, not a benchmark, and the article offers only Mayham's own explanation: "the intent signal is just different when someone is having a conversation with an AI and it recommends you by name." One hypothesis is that the user has already evaluated options inside the AI answer and chosen to click through for a specific reason. Measure your own AI-referred conversion against organic to see whether the same holds for you.
What is Machine Relations and how does it connect to zero-click search?
Machine Relations is the discipline of earning AI citations and recommendations for a brand by making it legible, retrievable, and credible inside AI-driven discovery systems. Machine Relations was coined by Jaxon Parrott, founder and CEO of AuthorityTech, in 2024. It connects to zero-click search because the mechanism is the same one PR has always used — earned media in trusted publications — now applied to the machine readers that answer buyer queries in place.
How can companies measure whether AI engines are citing their brand?
Start by querying ChatGPT, Perplexity, Gemini, Google AI Mode, and Claude with the buyer queries that matter most to your category. Document whether your brand appears, what position it holds, and what sources are cited. Track AI bot traffic (ChatGPT-User, PerplexityBot, OAI-SearchBot, ClaudeBot) in your analytics as a leading indicator of future citation. AuthorityTech's visibility audit gives you an AI visibility score instead.
Where do GEO and AEO fit in a zero-click pipeline strategy?
Generative Engine Optimization (GEO) and Answer Engine Optimization (AEO) are operational layers within Machine Relations. GEO optimizes content to be cited in AI-generated answers. AEO optimizes for featured snippets and answer boxes. Both depend on the same upstream input: earned authority in sources that AI engines trust. In the Machine Relations stack, GEO and AEO sit at the distribution layer — they work only when the foundation layers of earned authority and entity clarity are already in place.