Entity Mentions & Click Signals
A second Google patent explains how brand-new pages can rank without any click history — by inheriting the click behavior already associated with the entities they contain.
What you'll learn
- Understand patent US10303684B1 and the "search term–entity selection value" mechanism
- Learn how new pages can get a ranking boost through entity association
- Know why explicit entity annotation matters more than keyword density
- Apply the entity-first mindset to new page launches
The Cold-Start Problem
New pages have no click history, so Google has no direct behavioral signal to judge them by. Patent US10303684B1 describes a fix: instead of waiting for a page's own CTR data, Google aggregates the CTR that a given entity already generates across the pages that already contain it, and uses that aggregate to boost or demote the new page.
How the Mechanism Works
- Every indexed resource is annotated with the entities it discusses (a brand, a place, a product).
- For each search term, Google aggregates the historical CTR ("selection value") of all resources that share an entity relevant to that term.
- When a new resource appears containing that same entity, its ranking is adjusted using the aggregated CTR — even with zero click history of its own.
If dozens of existing "Mount Fuji hiking" pages already show a high CTR for the query "climb Fuji," a brand-new article that explicitly discusses the entity "Mount Fuji" can inherit part of that boost before it has any clicks of its own.
Why Explicit Entity Annotation Matters
This system only works if Google can reliably detect which entities a page is about. That means: naming entities explicitly in the text (not "it," "this brand"), using structured data (schema.org) to declare entity types, and staying consistent about entity names across the page.
Practical Application for New Pages
When launching a new page in a competitive topic, prioritize clearly naming the strong, well-established entities relevant to the page (brands, products, locations) rather than relying purely on generic keyword phrasing. New product listings, for instance, benefit from being explicitly tied to entities that already have strong CTR history in similar searches.
Key Takeaways
- Patent US10303684B1 describes ranking adjustment through aggregated entity-level CTR, not just page-level CTR
- New pages can inherit a ranking boost from entities they contain, even with zero click history of their own
- This only works if entities are explicitly and unambiguously identifiable in the content and markup
- Entity-first SEO (naming entities clearly, using schema types) is a direct, actionable consequence of this patent
- CTR by entity, not just by page, is a signal worth tracking in competitive research