iPullRank: keyword research now means managing portfolios

What happened

Per iPullRank, the keyword research playbook that defined SEO for two decades is collapsing under AI search. The agency's argument: a single phrase no longer maps to a single page or a single ranking. Instead, queries fan out into clusters of related prompts, model rewrites, and follow-up questions, each with its own retrieval logic and its own potential citation.

iPullRank reports that practitioners need to start treating keywords the way an investor treats holdings: as a portfolio with exposure, concentration risk, and correlated returns. The piece frames three shifts. Keywords are no longer atomic units; they are bundles of intent. Rankings are no longer the unit of measurement; citation share across LLMs is. And the page is no longer the unit of optimisation; the entity is.

The shift sounds semantic. It is operational. It changes how teams staff, brief, and measure content.

Why it matters for your brand

If you run content for a bank, an asset manager, or an insurer, the portfolio framing is already familiar, which is the point. You do not optimise a single fund; you optimise exposure across a book. The same logic now applies to how ChatGPT, Perplexity, Gemini, and Claude surface your firm in answers about ESG disclosure, private credit, or operational resilience. One winning page on "Basel III endgame" does not protect you. The model will rewrite the query thirty ways, and your citation share is the average across all of them. Financial services teams that still report on flat keyword rankings to their CMOs are measuring the wrong thing.

For multilaterals and UN agencies, the stakes are different but the structure is identical. When a policy researcher asks an LLM about disaster risk financing, climate adaptation metrics, or SDG progress, the model pulls from a basket of sources. UNDRR, the World Bank, OECD, and a long tail of academic papers compete for slots inside a single answer. The question for a communications lead is not "do we rank for this term." It is "what is our citation share across the cluster of prompts a policymaker would actually ask, and how concentrated is our exposure in any one model." Concentration risk in citations is a real failure mode. If 80% of your visibility comes from one Perplexity retrieval pattern, a model update can wipe it out overnight.

For major industrial groups, the portfolio view exposes an awkward truth: most B2B content programmes are wildly over-indexed on top-of-funnel explainer content and under-indexed on the technical, specification-level pages that LLMs actually cite when a procurement lead asks a substantive question. Holcim does not need to win "what is low-carbon cement." It needs to win the specification, standards, and case-study queries that decide a tender. That requires auditing the portfolio, not adding more blog posts.

For philanthropic and policy institutions, the implication is that thought leadership now has to be machine-legible at the entity level. The model needs to know what your organisation is, what it has published, and what it is authoritative on. Foundations that publish dense PDFs and call it a day will be invisible. The ones that structure their output as discrete, citable claims tied to named authors and clear entities will dominate the citation share for their issue area.

The content strategy implication across all four sectors: stop briefing writers on keywords. Start briefing them on prompts, intents, and the cluster of related questions a real reader would ask. Stop measuring rankings. Start measuring citation share, model by model, prompt cluster by prompt cluster. And staff for it. The skill set is closer to equity research than to SEO copywriting.

The signal in context

This is the third or fourth credible argument in the past six months that the unit of SEO measurement is broken. Other practitioners have made adjacent points: that AI Overviews compress the funnel, that brand mentions in training data matter more than backlinks, that entity authority outranks page authority. iPullRank's contribution is to give the shift a usable mental model. Portfolio management is something every senior marketer at a bank, an industrial group, or a multilateral already understands intuitively. Reframing keyword strategy in those terms is the kind of move that gets a CMO to fund the work.

The deeper trend is that AI search is forcing B2B marketing to behave more like equity research and less like demand generation. You are managing exposure to a set of model behaviours that you do not control, cannot fully observe, and that change without notice. The teams that win will be the ones who build the muscle to monitor, rebalance, and concentrate their bets deliberately. The ones who keep counting blue links will not see the decline until the pipeline shows it, and by then the rebuild takes a year.