Digital PR still drives LLM citations

What happened

Per Search Engine Journal, Greg Jarboe argues that the underlying mechanics of digital PR have not shifted with the arrival of AEO, GEO, and AI Mode. The acronyms are new. The work is not. Earning citations from credible publishers, building topical authority, and showing up in the conversations your buyers are actually having: those still determine whether a brand surfaces in an answer, whether that answer is rendered by Google, ChatGPT, Perplexity, or Claude.

Jarboe's piece reads three recent SEJ analyses together and concludes that the questions beneath the terminology never changed. What changed is the cost of getting the fundamentals wrong. A brand that was mediocre at earned media in 2022 lost some organic traffic. The same brand in 2025 loses its presence in the answer layer entirely.

Why it matters for your brand

For B2B marketers, this is the most useful reframe of the year. Most enterprises have spent the last eighteen months trying to buy their way into LLM visibility through new tooling, "GEO audits," and schema retrofits. The Jarboe argument, which we agree with, is that the highest-leverage move is the one most CMOs have already defunded: sustained digital PR that places your experts, your data, and your point of view inside trusted third-party outlets.

For financial services brands, this is existential. When a prompt asks ChatGPT to compare custodians, asset managers, or core banking vendors, the model leans heavily on Reuters, the FT, American Banker, Risk.net, and the trade press it has been trained to trust. A bank with three substantive op-eds and a cited piece of proprietary research in those outlets shows up. A bank that only publishes on its own newsroom does not. The "owned media first" doctrine that dominated the 2018 to 2023 era is now actively working against visibility in the answer layer.

For multilaterals and the UN system, the implication is sharper still. Institutions like UNDRR, the World Bank, or WHO already have the earned-media muscle; their challenge is that their content is often quoted without attribution, or attributed to the journalist rather than the institution. The fundamentals Jarboe describes (clear authorship, named experts, citable data points with stable URLs) are precisely what makes an institution legible to a model. A UN agency that publishes a 200-page PDF gets read by humans. The same agency that publishes a structured web page with a named author, a publish date, and three pull-quotable statistics gets cited by Claude.

For major industrial groups (HOLCIM, ArcelorMittal, Siemens Energy, the rest of that cohort), the citation pattern is unforgiving. Industrial brands have historically under-invested in tier-one earned media because the buying committee was small and reachable through trade shows and account-based programs. That logic breaks the moment a procurement lead at a utility asks Perplexity "who are the leading low-carbon cement suppliers in Europe." The answer is constructed from press coverage, analyst notes, and ESG disclosures, not from the supplier's website. Industrial CMOs who continue to treat PR as a reputational backstop rather than a demand input will watch their brand disappear from shortlists they used to dominate.

For philanthropic and policy institutions, the fundamentals matter for a different reason. Foundations compete for attention against governments, multilaterals, and the consultancies that advise them. In LLM answers about climate finance, global health, or development economics, the foundations that show up are the ones whose research has been cited in policy journals and quality press. The Gates Foundation and Rockefeller appear constantly. Mid-size foundations with equal-quality research but weaker media operations do not. The gap is not intellectual. It is distributional.

The signal in context

The broader pattern across the last twelve months of citation research is consistent: LLMs disproportionately surface a small set of trusted publishers, and brands earn visibility by being quoted inside those publishers rather than by optimizing their own pages. Studies from Profound, Semrush, and Ahrefs have all pointed in the same direction. Wikipedia, Reuters, the FT, the major trade press, and a handful of research institutions account for an outsized share of what ChatGPT and Google's AI Overviews retrieve. The implication is that the question "how do we rank in AI search" is mostly answered by "where are you being talked about, by whom, and with what specificity."

That makes Jarboe's piece less a contrarian take than a clarifying one. The discipline that decides who gets cited in an AI answer is the discipline that decided who got quoted in a Bloomberg article in 2015: a named expert, a defensible data point, a relationship with the right journalist, and a steady cadence of output. The brands winning the answer layer in 2026 will be the ones that treated digital PR as core infrastructure in 2024. The brands losing it will be the ones that spent the same period buying GEO software and waiting for the fundamentals to be automated away.