Google BERT
Also called: bert, bert update, bert seo
Google BERT (Bidirectional Encoder Representations from Transformers) is a natural-language model added to Google Search in 2019 that reads each word of a query in the context of the words around it, including small words like "to" and "for," to grasp intent instead of matching keywords in isolation.
How BERT changed query understanding
Before BERT, Search leaned heavily on keyword matching, which pushed people into typing “keyword-ese” instead of natural phrasing. BERT, built on the transformer architecture from the 2018 research paper by Devlin and colleagues, processes each word bidirectionally, in relation to everything that comes before and after it in the sentence. In Google’s own example, that lets Search register that “to” in “2019 brazil traveler to usa need a visa” means a Brazilian traveling to the US, not the reverse.
At launch in October 2019, Google said BERT would help it better understand about one in ten English searches in the US, and it also applied a BERT model to featured snippets across two dozen countries. Google expanded it to more than 70 languages that December, and at its Search On 2020 event said BERT was used on almost every English query. It works hardest on longer, conversational queries where small connecting words shift meaning.
BERT is not a penalty and not something you optimize against. There is no “BERT keyword” to chase. It rewards content written the way people actually phrase questions, and its context-reading approach has since been joined by later systems like MUM and the generative models behind AI Overviews.
How it affects your traffic
Because BERT reads the surrounding context of a query, padding pages with keyword-ese does nothing; pages that answer a real question in plain language win the passages BERT rewards. The same context-reading that powers BERT now underpins AI Overviews, ChatGPT, and Perplexity, which lift clear, self-contained answers straight into their responses. Writing for how people actually ask, and structuring each page so a single passage can stand on its own, is the core of the AI SEO work we do.
Get AI SEO that moves the needle
We turn terms like this into ranked pages and qualified pipeline. Start with a free Initial SEO Strategy.