Google’s Knowledge Graph Prunes 3 Billion Entities, Shifting SEO and AI Visibility
Build a clean, consistent entity definition across owned and earned media to strengthen your brand’s presence in knowledge graphs and future context graphs.
Audit and standardize your brand’s schema markup across all owned properties; update Wikidata and review platforms with consistent identifiers; document entity definitions for internal use; coordinate with data teams to populate context graphs.
Summary
Google’s Knowledge Graph, a vast knowledge base that contains 1.6 trillion facts about 54 billion distinct entities, has long been the engine behind the search engine’s rich results and its newer AI‑generated answers. By moving beyond simple keyword matching, the graph allows Google to interpret intent and surface relevant information even when a user’s query is phrased in an unexpected way. For example, a search for “small green guy with lightsaber” can return a precise answer because the graph recognizes the underlying entity and its attributes.
In June 2025, Google announced a sweeping pruning operation that removed more than three billion entities in a single week. The company said the move was intended to streamline the dataset for its AI products, making the remaining signals higher‑quality and more reliable. The pruning has increased the value of each entity signal, meaning that brands that do not maintain consistent, structured data risk disappearing from Knowledge Panels, AI Overviews, and the responses generated by Gemini. The change also signals a shift toward a tighter integration of the Knowledge Graph with Google’s AI Mode, which highlights connected entities in the chat interface and lets users click into panels for deeper context.
The pruning has tangible consequences for search visibility. Roughly 60 % of searches now end without a click because the Knowledge Graph or AI Overviews answer the query directly. This trend has reduced organic click‑through rates for many sites. To remain visible, the article recommends that marketers verify their entity recognition through the Knowledge Graph API, strengthen public‑relations and link‑building efforts, and validate schema.org Organization markup on their homepages—specifically the name, logo, URL, sameAs, and @id fields—using tools such as Ahrefs Site Audit.
For brands, the lesson is clear: structured data is no longer optional. As Google tightens its data curation, those who invest in accurate schema markup and robust link profiles will be better positioned to maintain or grow their presence in both traditional SERP features and the emerging AI‑driven answer ecosystem.
Key changes
- Retrieval layer uses RAG to fetch content chunks; limited reasoning.
- Knowledge graph (Google, Microsoft, Wikidata) defines brand entities.
- Schema markup and consistent naming improve entity recognition.
- Context graph incorporates governance, policies, and real‑time data.
- Google Cloud Next ’26 launched Knowledge Catalog for dynamic context graphs.
- Enterprise agents will use context graphs for internal decision making.
- Marketing teams must ensure entity consistency to avoid fuzzy recognition.
- Retrieval alone cannot guarantee AI visibility; higher layers are critical.