By David Carrasco Pamies and adapted for The Business Brief readers
Introduction
International keyword research can look precise and scientific, but a single early choice can quietly steer the whole project. In a practitioner study published on Search Engine Land, David Carrasco Pamies shows that the very first seed phrase used in a keyword tool can determine whether a market appears at all or remains invisible to analysts and executives. See the original piece on Search Engine Land for full detail. https://searchengineland.com/keyword-seed-selection-international-seo-488362
Why the first seed is a business assumption
It is tempting to treat the initial keyword as an administrative step – a translation task before the “real” research begins. Pamies demonstrates that this initial string is an implicit theory about how a market names itself. In regulated and qualification-led sectors that theory is especially risky. Descriptive labels can miss the institutional shorthand people actually use, such as Part 107 in the United States, AESA A1 A3 in Spain, EPA 608 for refrigerant handling, or industry acronyms like NMLS. When a tool never constructs a neighbourhood around the neutral seed, a market may appear to have no demand even though searches and commercial needs exist under different vocabularies.
The experiment and what it showed
Pamies ran a controlled comparison on August 21, 2026, across two keyword providers, Semrush and DataForSEO, using neutral descriptive seeds, local occupational terms, and verified institutional artifacts. The experiment was deliberately small and practitioner focused, but it is instructive. Semrush frequently produced empty result sets from neutral seeds – an entry failure. DataForSEO more often returned populated sets that nonetheless omitted the institutional artifact – a discovery failure. The two providers disagreed about which cases they recovered, making provider choice a visible methodological variable rather than a hidden assumption.
Different failure modes, different remedies
Treating every thin or empty export the same is a mistake. Pamies separates four diagnostic outcomes and maps each to a practical response:
– Entry failure: the seed returns nothing. Test other seed families, inspect the SERP, and run the concept through a second provider before concluding zero demand.
– Thin entry: a handful of rows with weak volume. Augment the keyword evidence with non-search sources such as job listings, support tickets, sales queries, and regulator pages.
– Discovery failure: a large-looking neighbourhood that nonetheless omits the controlling credential or exam. Reverse-seed from the artifact to surface the missing needs.
– Provider disagreement: different tools build different markets. Keep provider provenance attached to exports rather than averaging blind outputs.
Build a market-entry map before you open the tool
Pamies offers a straightforward alternative to the translate-and-run habit. Before any tool queries, construct a market-entry map listing four seed families and recording their provenance:
– Neutral description: the outsider translation or plain-language category label.
– Local occupational term: how workers, employers, and customers name the role in practice.
– Institutional artifact: the license, exam, card, code, or statutory title that gates access.
– Commercial or legacy term: brands, historical credentials, or market labels still in use.
Record where each string came from – regulator pages, job boards, competitor copy, internal search logs, or practitioner interviews – and run these families separately in the keyword tool. Only after you know which door opened the market should you consolidate terms for clustering and modelling.
Reverse seeding reshapes content architecture
One important consequence beyond keyword lists is content design. An occupational page describing a service or course is not the same as a hub built around an institutional artifact. Reverse seeding often reveals distinct user journeys: eligibility, exam preparation, renewal, documentation needs, and region-specific rules. Those needs justify separate hub pages and different internal linking than a single category page would imply.
Practical rules for teams and decision makers
Pamies proposes simple process safeguards that are easy to adopt and materially change outcomes:
– No zero in a market-sizing deck until at least one alternative seed family, one non-keyword evidence source, and one second retrieval route have been checked.
– Preserve seed family, seed provenance, provider, retrieval route, and diagnostic outcome in every export.
– Ask local reviewers to validate entry maps, not only translations, before large-scale expansion.
– Use reverse seeding from known artifacts to expand content plans and to test whether descriptive routes missed high-intent needs.
What this means for Irish businesses
For Irish companies and agencies expanding internationally, the lesson is immediate and practical. Markets shaped by regulation, professional licensing, or public procurement will often use local shorthand that English or translated Spanish will not reveal. A neutral description can undercount opportunity, and a careless zero can shift budgets away from viable markets. Investing a little time up front to map local vocabularies, read regulator guidance, and consult practitioners will usually save money and avoid strategic missteps.
Tools matter, but so does method
Pamies is careful to note the study is small and not a general statistical claim. It measures tool outputs rather than searcher behaviour, and it focuses on retrieval logic rather than on Google SERP rewrites. Still, the structural point is robust: seed dependence is a feature of keyword research. Different providers encode different index and retrieval theories, and teams should surface those theories rather than hide them in a finished spreadsheet. For teams wanting to test providers, Semrush and DataForSEO are useful starting points. https://www.semrush.com https://docs.dataforseo.com
Takeaway and an operational checklist
– Before the tool: build a market-entry map with four seed families and record sources.
– Inside the tool: run seed families separately and label every export with provider and retrieval route.
– After the tool: reverse-seed institutional artifacts, compare journeys, and translate findings into page types and internal linking.
– Governance rule: do not accept a zero until you have validated at least one alternate seed, one non-keyword evidence source, and one second provider.
Conclusion
Keyword research is not just a technical task. The first string you type expresses a business assumption that can determine what your models see and what they ignore. David Carrasco Pamies exposes that fragility and offers an operational fix that is modest in effort and large in consequence. Read the full analysis on Search Engine Land to follow the methodology and case examples in depth. https://searchengineland.com/keyword-seed-selection-international-seo-488362
