Search Grounding (or Grounding with Search) is the mechanism by which generative AIs run real-time web searches to supplement their answers. Instead of relying solely on training knowledge, the AI consults current sources before responding.
This is extremely relevant for GEO because it means recent, indexed content has a chance of being cited immediately, without waiting for the model's next training cycle. If you publish an article today and Google indexes it, an AI with search grounding can cite it tomorrow.
In practice, to take advantage of search grounding, your content needs to be indexed quickly (an up-to-date sitemap, Google Search Console configured), be relevant to the query and come from a trusted source. It is one more reason why SEO and GEO are complementary: SEO ensures fast, proper indexing, and GEO ensures the content is quotable.