In early 2025, LinkedIn’s organic growth team had a problem that wasn’t showing up in their rankings.
Non-brand, awareness-driven traffic had dropped by up to 60% across a subset of B2B topics — while keyword positions held. Their pages were still ranking. Clicks were just disappearing. Google AI Overviews had started answering the informational queries LinkedIn ranked for, serving the summary directly without sending the click.
They’d been tracking this since Google first rolled out Search Generative Experience in early 2024. By early 2025, the scale of the impact was hard to ignore. Their response is worth studying: they didn’t hire more SEO specialists.
The reflex for most marketing teams facing an organic traffic decline is to run an SEO audit — fix technical issues, improve content quality, wait for recovery.
That works when rankings are the bottleneck. LinkedIn’s rankings weren’t the bottleneck.
AI Overviews appear in roughly 30% of search results and reduce click-through rates by 35% when present, according to large-scale Similarweb analysis published by Search Engine Land. The page ranks. It might even get cited inside the AI summary. But the click goes to zero because the AI absorbed the intent before the user needed to leave.
Wynter’s 2026 survey of B2B CMOs puts harder numbers on what this means for the funnel: 84% of B2B buyers now use AI for vendor discovery, and 68% start their search in AI tools before they open Google. They narrow down options in ChatGPT, then use Google to verify. If your brand doesn’t appear in the AI answer — if the AI doesn’t cite or recommend you when your category comes up — you’re invisible during the evaluation phase. Whether you rank number one or not.
This is the gap LinkedIn hit. Ranking first doesn’t get you on the shortlist if the shortlist is built before Google gets involved.
LinkedIn’s solution was cross-functional. Their AI Search Taskforce spanned eight functions: SEO, PR, editorial, product marketing, product, paid media, social, and brand.
Here’s what each one actually does in this context.
SEO continued doing what it was already doing — semantic structure, clear heading hierarchy, technical fundamentals. This is still necessary. AI systems need to find and parse your content. But it’s the floor, not the lever. LinkedIn didn’t reach the second most-cited domain in Google AI Mode by improving H2 tags.
PR and editorial was the function that produced measurable citation lift. LinkedIn noted “a meaningful lift in visibility and citations” specifically from earned content and editorial placements. Third-party coverage in publications AI engines weight as authoritative builds a citation layer that on-page optimization can’t create. This is the function most B2B teams have cut or deprioritized — and the one where the gap shows up most directly in citation frequency.
Product marketing focused on accuracy: correcting AI systems when they describe what you do wrong. If ChatGPT explains your product incorrectly, no volume of content optimization fixes it. Product marketing’s job is to systematically build the correct description into every surface AI systems might encounter — third-party coverage, review sites, analyst write-ups, and directory listings.
Product ensured that internal search pages and structured surfaces were properly crawlable and indexed. Separate data on B2B SaaS sites from Search Engine Land’s AI traffic analysis shows AI penetration on search pages running 8.7x higher than site-wide averages. If those pages aren’t structured for retrieval, that surface area isn’t in play.
Paid media redirected spend toward branded and high-intent queries as informational awareness traffic migrated to AI answers. The economics of informational paid search shifted; budget concentrates on queries where the AI answer doesn’t fully resolve the intent, and where showing up in the post-AI verification step still matters.
Social — including LinkedIn’s own platform — tested owned content for AI discovery performance. The results: according to Semrush’s analysis of AI citation patterns from November 2025, Google AI Mode now cites LinkedIn in roughly 15% of responses, making it the second most-cited domain behind YouTube. Platforms AI engines actively pull from have structural advantages worth building on deliberately.
Brand covered monitoring (tracking what AI systems say about you), correcting misinformation when it appears in AI responses, and building entity coverage through podcast appearances, customer reviews, and forum presence. If a brand appears alongside a category repeatedly across the web, AI systems recognize it as a category authority over time. This is less about any single campaign and more about coverage density across sources the AI trusts.
Of the eight, PR and editorial is where most B2B marketing teams have fallen behind. The agency relationship produces one placement per quarter. The shift toward owned content over the past few years made earned coverage feel less urgent. Budget went elsewhere.
That’s the gap that shows up directly in citation scores.
AI engines cite publications, not brand websites. AuthorityTech’s research on AI citation patterns across more than 1 million citations in 11 industries shows a clear tier structure: Forbes, Reuters, the Financial Times, and Axios appear consistently across B2B sectors as the first tier for universal AI visibility. These are the publications where your category competitors either do or don’t have coverage. If they’re there and you’re not, the AI’s answer to “who are the top [your category] companies” reflects that gap.
The volume and quality of earned placements in publications AI engines trust is the primary driver of citation frequency. This isn’t a content quality problem. It’s an editorial relationship problem — one that requires a different kind of investment than most SEO roadmaps account for.
LinkedIn’s new operating model — “be seen, be mentioned, be considered, be chosen” — is a practical description of how AI-era discovery works. Being seen means appearing in publications AI engines cite. Being mentioned means your brand surfaces in AI answers. Being considered and chosen follow from that.
This is what Machine Relations formalizes as a discipline: earned media placements in trusted publications create the citation infrastructure that drives AI recommendations, which drives the pipeline that used to come from informational search traffic. The mechanism is the same one that made PR valuable in the first place. The reader changed from human to machine.
LinkedIn had the traffic data to force the org restructuring. Most B2B companies are still waiting for that data point before they act.
The audit that shows you where you stand — before you’re down 60% — is at app.authoritytech.io/visibility-audit.