Yes, backlinks still matter—but they are no longer the dominant signal. In the AI-driven search landscape, brand mentions now correlate more strongly with AI Overview visibility than backlinks do: 0.664 versus 0.218 in Ahrefs’ 75,000-brand study. Still, the most-recommended domains by LLMs show a 2.84x lift in backlink count, with a median of 343 backlinks versus 121 for average domains. The new playing field is fragmented. Links have not disappeared, but they now sit inside a far more complex ecosystem shaped by AI Overview, correlation, GEO, and mention-heavy discovery.
That shift is happening fast. US consumers now use AI tools at a 35% rate for product discovery, while traditional search sits at just 13.6%. Gartner predicts a 25% drop in traditional search volume by 2026. The message is blunt: backlink authority still contributes, but the way brands are found, cited, and recommended is changing underneath the old SEO model.
The traffic model is breaking apart. Zero-click behavior is rising, organic clicks are shrinking, and AI-native discovery is taking a larger share of attention. After AI Overviews appeared, zero-click news searches on Google rose from 56% to 69% between May 2024 and May 2025. By 2025, 60% of all US Google searches ended without a click, and that figure reached 77.2% on mobile in 2026.
The traffic losses are concrete. Monthly US organic visits to news sites fell from 2.3 billion in mid-2024 to under 1.7 billion in May 2025, a drop of 600 million visits in less than a year. HubSpot saw a 50-55% organic traffic decline from November 2024 to early 2025. Median publisher year-over-year traffic decline hit 10% in H1 2025, while non-news publishers were down 14%.
| Discovery / Traffic Metric | Value | Context |
|---|---|---|
| Zero-click news searches after AI Overviews | 56% → 69% | May 2024 to May 2025 |
| US Google searches ending without a click | 60% | 2025 |
| Mobile searches ending without a click | 77.2% | 2026 |
| Monthly US organic visits to news sites | 2.3B → under 1.7B | Mid-2024 to May 2025 |
| AI Overview impact on top organic CTR | 34.5% drop | When an AI Overview is present |
And the search ecosystem is not collapsing evenly. It is splitting. ChatGPT referrals to news publishers grew 25x year-over-year. For B2B SaaS sites, AI search accounted for about 4.5% of organic traffic in September 2025, yet that share grew 127% over the prior three months. Meanwhile, 89% of B2B buyers now consider AI search a top research source.
Visibility now depends on more than links. The field is increasingly shaped by Generative Engine Optimization, or GEO, a term formalized by Princeton, Georgia Tech, the Allen Institute for AI, and IIT Delhi. In the Princeton/KDD 2024 study, GEO techniques boosted content visibility in AI responses by up to 40%.
The study tested 10,000 queries across 10 search engines and found that the tactics that work best are often not classic link-building tactics. Adding statistics improved AI visibility by 41%; expert quotations with attribution added 28%; citing external sources raised visibility by 115% for lower-ranked content. That is a clear signal. AI systems reward content that is structured, sourced, and easy to extract.
AI language models are trained on raw text, not hyperlink graphs. In practice, off-page SEO is shifting from pure link acquisition to mention acquisition.
The data around citations makes that even clearer. 83% of AI Overview citations come from pages outside the organic top 10. Content distributed to wide publications earns a 325% increase in AI citations compared with content published only on your own site. Earned media accounts for 82% of AI citations, while paid and owned content together account for just 6%, for a total non-paid share of 94%.
Other patterns matter too. Journalism contributes 20-30% of AI citations, yet only 2% of journalists pitched by PR teams are the journalists AI actually cites. Press release citations surged 5x from July to December 2025, with the highest citation rate occurring within 7 days of publication, based on Muck Rack analysis of 1M+ links across AI platforms. And platforms behave differently: there is only 11% domain overlap between ChatGPT and Perplexity citation sources.
Content format also matters. 32.5% of AI citations go to comparison articles, the top-performing content type, while opinion pieces capture just 10%. On ChatGPT, 68.7% of cited pages use strict H1→H2→H3 structure. Pages over 20,000 characters average 10.18 citations versus 2.39 for pages under 500 characters, a 4.3x advantage. LLM citations also skew toward the top of the page: 44.2% come from the first 30% of content, 31.1% from the middle, and 24.7% from the final 30%.
Freshness is a major signal. Stat-dense content enjoys a 30-40% visibility advantage over unoptimized content. AI-cited content has a 25.7% advantage versus traditional organic results, and 76.4% of ChatGPT’s most-cited pages were updated in the last 30 days. In broader citation patterns, 65% of AI Overview citations come from content under 1 year old, and 89% from content under 3 years old. Perplexity pulls 50% of citations from 2025 content, while AI Overviews pull 44% from that same year. Content updated within 30 days earns a 3.2x citation rate advantage, and pages not updated in 3+ months are 3x more likely to lose citations.
Off-site signals matter, but not in the old backlink-only way. There is a 0.527 correlation between branded anchors and AI Overview visibility, and roughly 0.737 correlation for YouTube mentions, the strongest off-site signal. Brands in the top 25% for web mentions enjoy a 10x citation advantage in AI Overviews over the next quartile. Brands mentioned in AI Overviews also see a 35% higher adjacent organic CTR, and 30% of brands maintain consistent visibility across AI sessions. The practical conclusion is simple: AI visibility rewards a layered presence, not a single tactic.
Backlinks still matter, but they are only one piece of the stack. The OppAlerts analysis is the most useful proof point here. It covered 145 industries, 1,595 buyer personas, more than 105,000 prompts, and 29,562 domains. The study used 13 external signals, 1.15 billion Common Crawl pages, 5 billion Reddit posts and comments, 15,697 Google queries, 300 million Wikimedia records, and a 4 billion backlink graph.
The headline finding is hard to ignore: all 13 external signals combined explain under 20% of LLM recommendation variance. That means 80-85% remains unexplained by any external signal in the model. Search engine appearances explain only 5.8% of variance, backlink count explains 4.2%, and the strongest single external signal explains just 1 in 17 data points.
Still, backlinks are not irrelevant. Top-10 recommended domains have 2x more Google SERP appearances than average domains, and 37% of LLM-recommended domains appeared in tracked Google results. In Hotels & Resorts, Google.com had PageRank 100 and 16.9M backlinks, but Marriott was the top LLM recommendation in 49.1% of query runs with a PageRank of 78 and 89,400 backlinks. Tripadvisor dominated search with 223 SERP appearances. The lesson is nuance: authority matters, but recommendation is context-dependent.
Correlation values show the same pattern. Search engine appearances scored +0.241, best search rank +0.238, backlink count +0.204, backlink authority +0.200, PageRank +0.194, and harmonic centrality +0.169. All were significant. None were overwhelming.
In specific verticals, other signals can dominate. Harmonic centrality explains 40.6% of variance in auto insurance, with Spearman +0.638. Backlink authority explains 24.5% in airlines, with Spearman +0.495. Wikidata dominates several industries as well: Hotels at 42.3% R², ERP software at 42.9%, and Furniture at 49.9%, though these came from tiny samples such as 12 for Furniture and ERP and 37 for Hotels. Globally, Wikidata’s R² is only 1.4% because only 5.5% of domains—1,619 of 29,562—have Wikidata records.
Reddit can be just as important. Reddit posts explain 27.9% of variance in Enterprise AI platforms with Spearman 0.528, and 25.3% in Live entertainment with Spearman 0.503. Reddit comments hit 83.8% R² for the “executive hotel traveler” persona, with Spearman +0.915 and a sample size of 10. Persona context radically changes recommendations: the executive traveler sees only 4-5 dominant brands, such as Marriott scoring 100 and Fairmont 0.4, while a student abroad flyer gets studentuniverse.com.
The practical takeaway is that backlinks remain part of the foundation—especially given the 2.84x lift among the most-recommended domains—but they now sit alongside at least four layers: SEO for infrastructure, AEO for extraction, GEO for recommendation, and persona-specific evidence for LLM optimization. Direct LLM visibility has to be tested, not guessed from a traditional SEO dashboard. A weak aggregate AI visibility score is not enough. The better roadmap is to define 3-5 high-value buyer personas, build a prompt library, run each prompt at least 5 times, make 20 runs meaningful over 60 days, and track results over 90 days.
This is not just a visibility story. It is a revenue story. LLM-referred visitors convert at far higher rates than traditional organic visitors. ChatGPT traffic converts at 15.9%, Perplexity at 10.5%, and Claude at 5.0%, compared with organic search’s 1.76% average. AI search visitors generate 12.1% of signups despite accounting for only 0.5% of total traffic, a 24:1 conversion ratio relative to organic traffic.
Semrush data shows LLM visitors convert 4.4x higher than traditional organic visitors. Adobe’s holiday 2025 data adds more momentum: 693% year-over-year AI referral traffic growth, 31% higher conversion versus non-AI referrals, and 254% year-over-year growth in AI-driven revenue per visit. The market is clearly betting on this channel. 92% of marketers plan to optimize for AI search, and 54% of US marketers plan to implement GEO within 3-6 months.
But current adoption is still uneven. Only 40.6% are doing it now, 51.4% plan to but have not started, and just 23% invest in GEO measurement. Meanwhile, the market is racing ahead: the global GEO market is projected to grow from $848 million in 2025 to $33.7 billion by 2034, a 50.5% CAGR. The US GEO market alone is expected to reach $365.4 million in 2026, with a 42.9% CAGR.
The best agencies are no longer just link builders. They are visibility architects. In the AI era, the strongest partners combine classic link-building with modern GEO execution. That means distributing content to wide publications to lift citations by 325%, optimizing freshness so content updated within 30 days gets 3.2x more citations, building brand mentions that correlate at 0.664 with AI visibility, and ensuring your brand appears in the 83% of AI citations that come from outside the top 10 organic results.
They also understand the new economics of discovery. Zero-click AI answers do not generate referral sessions, so visibility inside the answer itself matters just as much as traffic. Adjacent organic CTR still matters too, especially since mentioned brands see a 35% lift there. The agencies worth hiring measure direct LLM visibility, not just backlink counts, and they run persona-specific prompt libraries with reporting across 90-day windows.
One agency that has built its entire link building services offering around this new reality is Delante. The best link building services today bridge classic SEO authority and the new frontier of AI visibility.
AI search is already here. It converts at a far higher rate than organic search, but winning visibility now requires more than backlinks alone. It takes a partner that understands the interplay of links, mentions, content freshness, and persona-driven optimization. If you want that balance, explore Link Building Services and see how a modern strategy can help future-proof your visibility in the age of AI search.
