Here is the Google BERT update explained in one line: BERT helps Google read a search the way a person would, with each word in the context of the whole sentence. Google announced it on October 25, 2019, for about one in 10 English searches in the U.S.
Google BERT update explained in plain words
BERT stands for Bidirectional Encoder Representations from Transformers. That sounds hard, but the idea is simple. BERT is a natural language processing (NLP) model. NLP is the field of teaching computers to understand human language.
Older systems read a query more or less word by word. BERT looks at the words before and after each word at the same time. That is what "bidirectional" means. So small words like "to", "for" and "no" now change the meaning, as they should.
What did BERT change in search?
Google said BERT would help Search "better understand one in 10 searches in the U.S. in English". Think of a search like "brazil traveler to usa need a visa". The word "to" matters a lot. The person is a Brazilian going to the USA, not the other way around. A word-by-word reading can miss that detail.
- Longer queries. BERT helps most with long, chatty searches.
- Small words. Prepositions and word order now count.
- Featured snippets. Google also used BERT to pick featured snippets in about two dozen countries.
Who lost or gained traffic?
BERT did not punish sites. It made Google better at picking the page that fits the question. That still moved traffic around.
- Gainers: pages that answer a specific question in clear, natural sentences. A travel blog post that explains visa rules for one country pair is a good example.
- Losers: pages that ranked because they contained the right keywords but answered a different question. Broad "ultimate guides" lost some long queries to focused pages.
Many site owners saw no big change. At first, BERT mainly changed long, specific searches.
Is BERT still running today?
Yes. BERT is active and Google lists it in its ranking systems guide. On December 9, 2019, Google said BERT was "now rolling out to over 70 languages worldwide." By October 2020, Google said BERT powered almost every English query. It works next to RankBrain and other systems.
This matters for small sites in other languages too. If you write in Spanish, Indonesian or German, BERT reads your pages with the same care for context. Clear writing in your own language helps just as much as it does in English. You do not need to write in a special style for Google.
How to check if BERT affected your site
- Open the Performance report. In Google Search Console, go to Performance and set a range from September to December 2019.
- Compare before and after. Use the Compare tab to set the four weeks before October 21, 2019, against the four weeks after it.
- Filter long queries. Use a query filter or export the data. Look at searches with five or more words, since BERT changed those most.
- Read the questions. For queries you lost, ask if your page really answers that exact question. For new queries, check that the page fits.
- Rule out other causes. Check the Manual actions report and any site changes you made in the same weeks.
What to do now
You cannot optimize for BERT directly. There is no tag or setting for it. But you can write in a way that BERT reads well:
- Write for people. Use full, natural sentences. Do not cut words out to fit a keyword.
- Answer one question per section. Put the question in a heading and the direct answer just below it.
- Be specific. Name the place, the product or the situation. Details help Google match your page to the right search.
- Aim for snippets. A short, clear answer can win the box at the top of results. The featured snippet cheat shows the format.
That is the Google BERT update explained: Google got better at reading language, so clear writing that answers real questions now wins more often.
Frequently asked questions
Can I optimize my content for BERT?
Not in a technical way. BERT helps Google understand queries and text. The best approach is clear writing that answers a specific question in natural language.
Does BERT work in languages other than English?
Yes. Google launched it for U.S. English searches in October 2019. In December 2019, Google said BERT was rolling out to over 70 languages.
Did BERT replace RankBrain?
No. Google lists both as active systems. They do different jobs and work together. RankBrain links words to concepts, while BERT reads words in context.