Wikidata

Wikidata is a free, collaborative, multilingual knowledge base run by the Wikimedia movement that stores facts about entities as structured, machine-readable statements. Every entity gets an item with a permanent ID (Douglas Adams is Q42), and statements connect that item to values, sources and identifiers in other databases. For search engines and AI systems, Wikidata works as a shared identity layer: a stable way to say which 'Mercury' or which 'Apex' a name refers to.

Wikimedia Deutschland launched Wikidata on October 29, 2012, initially to centralize the links between language editions of Wikipedia. All of its data is released under CC0, so anyone can reuse it, including commercially, without permission. In October 2026 its statistics page reported more than 123 million items and over 2.5 billion edits since launch.

An item has a few parts that matter for brands:

• A label, a short description and aliases, in as many languages as editors add.
• Statements made of a property and a value, such as official website (P856), founding date, headquarters location, industry or founder, ideally each with a reference.
• Identifiers that link the item to external databases and profiles, such as social media handles or library authority records.
• Sitelinks to the corresponding Wikipedia articles, if any exist.

Wikidata's connection to Google is historical as much as technical. Google ran Freebase, a structured database that helped feed its Knowledge Graph, and in December 2014 announced it would wind Freebase down and help move its data into Wikidata. Google does not publish a list of Knowledge Graph sources; its help center says facts come from a variety of public sources, licensed data and content owners. So a Wikidata item is best seen as one consistent, widely reused reference point, not a switch that creates a knowledge panel.

Wikidata is also becoming easier for AI systems to use directly. On October 1, 2025, Wikimedia Deutschland launched the Wikidata Embedding Project, a vector database built with Jina AI and DataStax, together with a Model Context Protocol server that lets AI agents search items and run queries. The project's stated aim includes reducing reliance on what models memorize by grounding answers in referenced statements.

Inclusion has rules. Under Wikidata's notability policy, an item is acceptable if it has a valid sitelink to a Wikimedia page, refers to a clearly identifiable entity that can be described using serious and publicly available references, or fills a structural need for other items. A company item supported only by its own website is likely to be nominated for deletion, and the community makes the final call.

For entity SEO, the practical steps are modest. Check whether your organization already has an item and that it is the right entity, not a namesake. Make sure the official website, founding date, location and industry are correct and referenced to independent sources. Then add the item's URL to the sameAs list in your Organization schema, which Google's structured data documentation describes as URLs of pages on other sites with more information about your organization, and keep it consistent with your entity home. None of this makes AI models recommend a brand. It reduces the chance that they confuse you with someone else or repeat outdated facts.