Hub and Spoke Planning Built from Live SERP Overlap

Of the topic clusters we map for an energy or fintech account in Spain, we recommend building about a third. The rest are named and refused with the reason written beside each: five of the top ten are comparadores owned by the same two groups, or the query returns a regulated-answer SERP where Banco de España, the CNMV and the OCU sit above everything commercial and will keep sitting there. The kill list is the part of this plan clients remember a year later. How the clusters are made: Not from text similarity. We pull the live top ten for every candidate query through DataForSEO in the correct locale, then group queries that share at least three ranking URLs. Two queries a keyword tool treats as near-identical often return different result sets in es-ES, and two that look unrelated return the same eight domains. The SERP decides what a single page can serve; hubs, spokes and the internal links between them follow from that grouping rather than from a column of search volumes. What a hub is in this practice: A URL that earns links from its spokes and passes intent downward. It is not defined by word count. Each hub is specified by the decision the reader is making at that moment, the queries it owns, the queries it must not target because a spoke owns them, and the modules the page needs: comparison table, eligibility criteria, calculator, the date the figures were last checked. Each spoke brief states the one question the page answers, the source of the number it has to contain, and the two links it sends, one up to the hub and one sideways to its sibling. Publishing order: Sequenced, hub live before its spokes, every spoke pointing at a hub that already exists. You get the order and the reasoning, not a calendar with thirty empty slots. A cluster finished properly outranks two clusters started, and output targets are the most reliable way a content programme dies. Where your own data enters: In these sectors the durable difference is a number nobody else holds: a dealer group's registration and residual-value history, an energy retailer's consumption curves by tariff, a lender's approval rates by segment. We check what you are legally free to publish, then design the cluster around it. If there is nothing publishable, we tell you before you commission a single word rather than after. Not included: Article production. We brief, we review against the brief and we reject drafts that missed it, but we do not staff a writing desk and we do not quote per article. Drafts written by a language model do not pass that review, not as a matter of principle but because they contain no fact that is not already ranking somewhere else. Who this is not for: Publishers who measure themselves in posts per month. Teams who want a keyword list with volumes attached; that already exists, it costs a hundred euros a month and this is not it. And anyone whose compliance function has not committed to a review turnaround, because in fintech and energy an unreviewed cluster is an unpublished cluster and the plan will stall at the same place every time. Languages and markets: es-ES and en, with es-MX and es-AR handled as separate SERP sets rather than one plan translated twice. Spanish is not a single market, and clustering in Madrid then republishing in Mexico City is the most common way a hub programme is wasted.