BERT stands for Bidirectional Encoder Representations from Transformers. Google began using it in search in 2019, describing it at the time as one of the biggest advances in the field's recent history.
BERT reads a sequence of words bidirectionally, considering the words either side of each term rather than processing left to right. This lets it interpret prepositions and word order that change meaning — the difference between travel to and travel from, for instance, which earlier systems frequently missed.
BERT marked the point at which conversational and complex queries began being understood properly. For content it reinforced a trend already underway: pages that genuinely answer a question rank better than pages that repeat its keywords, because the system now understands what the question means.
There is nothing to optimise for BERT directly, and Google said so explicitly at launch. Write clearly and naturally, answer questions completely, and avoid contorting sentences around keyword phrases. Clear prose is easier for language models to parse than keyword-optimised prose.
Want to learn more? Explore our SEO Glossary or check out our expert guides on SEO strategies.