LinkedIn AI Search Traffic Loss Hits 60% as Rankings Hold Steady

What LinkedIn actually disclosed

LinkedIn's marketing organization has published one of the clearest accounts yet of a problem many SEO teams already suspected. A page can keep its Google ranking and still lose most of its visitors. In a January 28, 2026 post on LinkedIn's marketing blog, the company's B2B Organic Growth team said non-brand, awareness-driven traffic fell by as much as 60% across a subset of B2B topics. Over the same period, rankings held steady and click-through rates softened.3 The team said it started hands-on testing of Google's Search Generative Experience, the beta that became AI Overviews, in early 2024. By early 2025 the effect on traffic was material and immediate.3

Search Engine Land's coverage put the numbers in context. It noted that LinkedIn never said how far click-through rates fell, and it pointed out that the effect showed up only once SGE turned into AI Overviews.1 The post was written by Inna Meklin, LinkedIn's Director of Digital Marketing, and Cassie Dell, Group Manager of Organic Growth. Their framing was that the old playbook still worked but needed updating.

LinkedIn's answer was organizational rather than purely technical. It set up an AI Search Taskforce drawn from SEO, PR, editorial, web marketing, product marketing, product, partner marketing, social, paid media and brand.4 It also dropped the "search, click, website" model in favor of a four-step sequence: be seen, be mentioned, be considered, be chosen.1

Read the fine print on the 60% figure

Much of the coverage reads the number in a way the original wording doesn't quite support. Headlines said LinkedIn "lost 60% of its traffic" or that its "B2B traffic dropped 60%."514 LinkedIn's own sentence starts with "across the industry," which mixes its own data with a wider market observation. It also limits the figure to "a subset of B2B topics."3 PPC Land quoted the "across the industry" phrasing directly.4 Other summaries described the figure as damage to LinkedIn's own properties.26

The best reading is that 60% is a ceiling for non-brand informational queries, not an average for LinkedIn as a whole. It is a meaningful warning sign. It is not proof that the network's search footprint collapsed. Several outlets listed what LinkedIn left out: which topics were measured, how big the sample was, what time period it covered, and how much the click-through rate actually fell.12 Some of the reporting also gets basic facts wrong. One summary put the SGE-to-AI-Overviews change in early 2026 rather than early 2025. Another dated the main announcement to February 10.12 That was actually the release date of a follow-up optimization guide. The strategy post came out on January 28.

The framework itself drew a mixed response. Search Engine Land said LinkedIn's content-level "learnings" mostly repeat established SEO and answer engine optimization (AEO) practice.11 That is a fair point. The real value of the disclosure is that a major brand put its name to the data, not that the tactics are new.

Rankings and traffic have come apart

For SEO practitioners, the important finding is that rankings and clicks no longer move together. For two decades the chain ran rank, click, visit, convert. LinkedIn's data shows the first link holding while the rest weakens.7 Outside studies point the same way. A SparkToro and Similarweb estimate found that about 60% of US and EU searches end without a click, and later reporting suggests that share is rising.4 Ahrefs data published in February 2026 linked AI Overviews to a 58% drop in click-through rate for results in position one.10 A randomized study reported in July 2026 found AI Overviews cut outbound organic clicks per search from 0.62 to 0.37.10

The losses are concentrated where B2B content marketing has always lived. One analysis noted that how-to guides, explainers and best-practice articles trigger AI Overviews most often. Those are the formats that drove B2B lead generation for the past decade.5 Branded and lower-funnel queries were not part of LinkedIn's reported losses. That fits the company's comment that its down-funnel performance held up early because of strong brand health.4

A new scoreboard, and a measurement gap

LinkedIn's operational changes say more than its slogans. The company says it now uses AI visibility software to track how its brand appears in LLM answers, in both branded and non-branded contexts. It has set KPIs around visibility rate, citation share and mentions across AI Overviews and LLM citations. It created a separate analytics traffic source for LLM-driven visits and monitors LLM bot activity through CMS logs.3 Its advice to other marketers is to start defining metrics such as LLM referral traffic, citation and mention volume, and AI Overview win rate now.4

The company is open about where this falls short. LLM-driven traffic to its B2B marketing sites grew by triple digits, and it can track how those visitors convert. But LinkedIn says it cannot measure how being visible inside LLM answers affects revenue, and it calls this the "dark" funnel.114 Observers pointed out that LLM referrals are still around 1% or less of total traffic for most sites. That makes triple-digit growth rates look larger than the volumes behind them.7 Figures for how well AI referrals convert vary a lot between reports. One cited study put AI search visitors at 4.4 times the conversion rate of traditional organic traffic.4 Another claimed 23 times.5 A gap that wide means the numbers depend heavily on how each study was designed, and they shouldn't be used as planning benchmarks.

What the playbook says, and how it has shifted

LinkedIn's early tests produced what it called a meaningful lift in visibility and citations. Owned content delivered the fastest and most scalable gains.4 The patterns it found are familiar to anyone who works in technical SEO. Clear headings and a logical information hierarchy help LLMs parse and surface content. The company also said LLMs favor material written by real experts, clearly time-stamped and written in a conversational, insight-driven style.4 A February follow-up guide, credited to AEO/GEO consultant Brooke Weller, argued for an "answer-first" approach instead of a keyword-first one. It recommends that each section stand on its own, because AI systems pull content out passage by passage.

A September 2026 account of the internal program described the changes in more detail. Adding FAQs to many pages reportedly produced a large lift in AI visibility. Schema, metadata and llms.txt improvements each added more. The team also consolidated a large number of underperforming URLs.8 Generative engine optimization (GEO) is starting to look like technical SEO with a different scoreboard.

By October, though, LinkedIn's messaging had changed. A LinkedIn marketing vice president argued that AI systems cite a distinctive point of view, not formatting.10 That doesn't fully contradict the earlier guidance. Structure helps a model find a passage, and originality gives it a reason to cite one. But the emphasis is clearly different.

The conflict of interest worth naming

All of this comes from a company with a clear stake in the outcome. Semrush data from November 2025 found Google's AI Mode cited LinkedIn in roughly 15% of responses, second only to YouTube.1 A larger Semrush study of 89,000 LinkedIn URLs in March 2026 again ranked it second among domains cited in AI answers.10 Later analysis found posts and articles made up 34.9% of LinkedIn's AI citations, while profile-page citations fell to 14.5%. PPC Land noted that LinkedIn sells advertising and gains directly when marketers conclude that publishing on LinkedIn is the route to AI visibility.10

That puts LinkedIn in an odd position. Its own marketing websites are losing clicks to AI Overviews, while its social platform has become one of the most-cited sources inside those same answers. One commentator summed it up: being the source an AI uses and being the page someone clicks are now separate outcomes.

The takeaway for SEO teams

The main lesson isn't the 60% figure. It's that LinkedIn stopped treating rankings as a stand-in for business impact and rebuilt its reporting around that change. Teams that still report mainly rank and organic sessions to leadership will find their numbers telling a less complete story every quarter. In practice that means:

• Keep doing technical basics like structure, schema, authorship and freshness.
• Start tracking AI citation share as its own metric.
• Bring PR, social and editorial into search planning.

LinkedIn's own measurement problem shows that no one has solved attribution for visibility that produces no click. The coverage agrees that visits are no longer enough to measure discovery. A reliable replacement metric doesn't exist yet.