BERT
Rollout of the BERT language-processing model, which understands the full context of a query by analyzing words in relation to one another. Google presents it as the biggest advance in five years.
Key points
- Rollout of the BERT language model to better understand the full context of queries.
- Analyzes words in relation to one another, not in isolation.
- Google presents it as its biggest advance in language understanding in five years.
- Impact concentrated on conversational and long-tail queries.
Analysis
BERT marks the shift from a keyword-based search to a fine understanding of intent. By grasping the role of each word in the sentence, notably prepositions and nuances, Google better interprets queries formulated in natural language, those that resemble a question asked to a human.
This advance foreshadows the logic of AI answers: the engine favors content that genuinely answers an intent rather than content that stacks up keywords. For visibility, writing clearly and answering questions precisely becomes more profitable than mechanical optimization.
What to do
- Write in natural language, explicitly answering users' real questions.
- Structure pages as questions and answers to capture the conversational long tail.
- Check for the presence of your content in AI answers and summaries on target queries.
Conversational and long-tail queries are better understood. A premium on natural content that genuinely answers the intent.