iPullRank: Google's AI search advice misleads marketers

What happened

Per iPullRank, Google's latest Search Central guidance on AI search is "naive and self-serving," and treating it as gospel will leave brands blind to how their content actually surfaces inside large language models. The agency's argument, published this week by founder Mike King's team, is that Google is telling marketers the AI search game is essentially still SEO, when the underlying retrieval and generation mechanics say otherwise.

iPullRank reports that two reflex camps have formed around every Search Central update: one screenshots a paragraph and declares "it's just SEO," the other declares SEO dead. Both are wrong, and Google benefits from the first camp staying loud. If marketers believe nothing has changed, they keep optimising for Google's index and stop investigating Perplexity, ChatGPT, Claude, and Gemini as distinct discovery surfaces with their own citation logic.

The core charge is that Google's public guidance describes a world in which classical ranking signals still govern AI Overviews and AI Mode. iPullRank's position: the retrieval layer, the grounding sources, and the way models select citations are materially different, and Google is not in a hurry to explain that.

Why it matters for your brand

If you run content or comms at a bank, an asset manager, a UN agency, an industrial group, or a foundation, the practical risk is that your SEO team is currently writing strategy decks based on Google's framing. That framing tells you to keep producing helpful content, keep your technical house in order, and trust that AI surfaces will reward the same signals. iPullRank's argument is that this is an incomplete map, and incomplete maps cost citation share.

For financial services brands, the stakes are immediate. ChatGPT and Perplexity are now the default research tool for a growing share of analysts, treasury teams, and corporate buyers. The sources those models cite when asked "who are the leading custodians for tokenised assets" or "which private bank has the strongest ESG framework" are not selected the way Google's blue links are. They are selected based on a mix of retrieval relevance, source authority signals the model was trained on, and, increasingly, real-time grounding from a small handful of partner indexes. If your SEO program is built on Google's guidance alone, you are optimising for one of those three layers.

For multilaterals and policy institutions, the asymmetry is worse. UN agencies, World Bank affiliates, and standards bodies have historically dominated the first page of Google for their topic areas because of domain authority and inbound links from governments and NGOs. That authority does not transfer cleanly into LLM citations. Models often cite secondary explainers, Wikipedia, and trade press over the primary institutional source, because those formats match the model's preferred answer shape. If a CGAP or an ISO assumes Google's guidance covers AI search, they will keep publishing 40-page PDFs and keep losing the citation to a Medium post that summarised them.

For major industrial groups, the issue is competitive intelligence inside procurement. Buyers asking ChatGPT to compare cement producers, logistics networks, or industrial automation vendors get a short list. The brands on that list are the ones whose product pages, sustainability reports, and trade press coverage are structured in ways the model can extract cleanly. Google's guidance does not tell you how to structure for extraction. iPullRank's broader point is that the gap between "ranks on Google" and "gets named in an AI answer" is now wide enough to be a board-level visibility question.

For philanthropic and policy institutions, the content strategy implication is sharper still. If your theory of change depends on your framing showing up in answers when a journalist, a policymaker, or a programme officer asks an AI for a briefing, you need to know which models cite you, which cite around you, and what the gap costs. Google will not tell you that. Your agency, your in-house team, or a dedicated GEO function has to.

The signal in context

iPullRank's critique lands at a moment when platform-issued guidance and independent measurement are pulling apart. Google has an obvious incentive to frame AI Overviews and AI Mode as continuous with traditional search, because that framing protects the SEO industrial complex that feeds Google's ad business. Independent operators, from iPullRank to Profound to Ahrefs, are publishing data showing that citation patterns inside ChatGPT and Perplexity diverge meaningfully from Google's top ten, and that the divergence is widening as model vendors strike their own content deals.

The takeaway for senior marketers is structural. Treating any single platform's public guidance as the source of truth for AI visibility is the same mistake as treating a single analyst report as the source of truth for a category. The brands that will hold citation share through 2026 are the ones running their own measurement across models, mapping where they are named and where they are not, and treating Google's Search Central documents as one input among several rather than the brief.