BERT

What is BERT?

BERT (Bidirectional Encoder Representations from Transformers) is a language model from Google that has been used in its search engine since 2019. It helps interpret the meaning of words in the context of the whole query, including small words such as “for” and “without”.

Why is BERT important?

• Natural language: Google understands long, conversational queries better.
• Fewer tricks: repeating keywords works worse than writing clearly.
• Context counts: the whole sentence decides which result fits.

Examples of BERT

• “can you pick up a prescription for someone else”: the small word “for” changes the question.
• “pizza without gluten delivery”: “without” decides which results fit.
• “bank” as a riverbank or a financial institution: the rest of the sentence gives the meaning.

How do you deal with BERT?

You cannot optimize for BERT specifically. Write naturally and completely, answer your reader’s question, and do not stuff sentences with repeated keywords. Your text then fits the way Google reads language. You can read more about writing copy under on-page SEO.