Grounding
Also called: grounded answers, grounding in sources
Definition
Grounding is the practice of anchoring a generated answer in specific sources: search results, documents, databases. The goal is to limit errors and to be able to show where each piece of information comes from. Google, for example, talks about grounding with Google Search for its Gemini models.
An engine that grounds its answers favors the sources it considers reliable and easy to use. Pages that state a fact clearly, with a number, a date or a name, are easier to pick up than vague promotional copy.
Example
An answer that says “the basic plan costs €11 a month, according to the pricing page” is grounded in a source. An answer that says “it's affordable” isn't.
What it means for your AI visibility
The more your pages give clear, verifiable facts, the more they serve as anchor points for answers. It's one of the reasons why pricing, documentation and FAQ pages get cited so often.
Related terms
- RAGRAG lets an AI look for documents before answering, then write its answer from what it found.
- CitationA citation is when an AI points to a page as a source of its answer, with a link. Unlike a plain mention, it can bring traffic.
- Citable contentCitable content is written so AI can reuse it as is: one idea per paragraph, precise facts, a direct answer to a question.
- HallucinationA hallucination is false information presented confidently by an AI: an invented price, a feature that doesn't exist, a fake quote.
Updated on · All glossary terms