By now, you’ve probably been inundated with commentary about Google’s BERT update – that’s Bidirectional Encoder Representations from Transformers, for the acronym lovers among us. So why another take?
Not to break down the name. Honestly, it sounds more like a sci-fi weapons system than an SEO tool. What matters is what BERT means for search – and for you.
Not Exactly Brand New
Let’s set the record straight: BERT isn’t a revolution. It’s a natural evolution.
This is Google continuing its mission to become the ultimate digital assistant – anticipating your needs, refining results, and serving up relevance at lightning speed.
Search used to be primitive. Early SEO meant obsessing over keyword density and hoping you’d nailed the exact phrasing users typed. It made for awful content and user frustration in equal measure.
To improve results (and hang onto users), Google had to get smarter. That meant stripping out spam, elevating quality, and developing a more nuanced understanding of how humans actually search.
Smarter Every Year
Google’s journey toward intelligent search has been underway for years. In 2013, Hummingbird ushered in semantic search – recognising that people don’t always use the same words for the same thing.
Then, in 2015, RankBrain introduced machine learning. Suddenly, Google was adapting on the fly: factoring in personalisation, location, search history, and context to interpret intent.
“User intent” became the cornerstone of modern SEO. You weren’t optimising for strings of text anymore – you were optimising for meaning.
The Role of BERT
BERT takes that idea further.
It helps Google better understand longer, more conversational search queries – especially when prepositions like “for”, “to”, or “with” carry real weight. Instead of parsing queries word by word, BERT reads them as wholes.
So, for example, a search like “2019 brazil traveller to usa need a visa” now leads to the right result: the U.S. Embassy in Brazil. Before BERT, Google might have assumed it was a U.S. citizen heading to Brazil and shown news articles instead.
That’s the difference: context-aware search that gets the nuance.
A Learning Machine
What makes BERT different from past updates is how it learns. Like RankBrain, it uses seed data and feedback from real-world searches – bounce rates, query reformulations, and patterns across billions of inputs – to constantly refine its results.
And it’s doing this 24/7.
Google’s own engineers admit they can’t fully explain how BERT works anymore. The system has grown too large and too dynamic to be reduced to a simple diagram.
Trying to manipulate it? You’re wasting your time. The algorithm now updates faster than any one person – or team – could possibly keep up with.
Who Should Be Worried?
Sites that write for algorithms, not people.
If your content is keyword-stuffed, vague, or written by AI with a loose grip on grammar, BERT is not your friend. But if you’ve been creating clear, useful, human-centred content all along? This is good news.
BERT is another step away from spam – and a giant leap towards content that actually helps people.
And while bad actors will always try to game the system, they’ll find fewer and fewer ways in.
Featured Snippets Get Smarter
BERT is also changing how Google chooses content for Featured Snippets – especially Position Zero.
Take a search like “parking hill no curb”. Previously, Google might have misunderstood and returned results about curbed parking. Now, it gets that “no curb” matters – and serves up guidance that matches the actual query.
So far, BERT hasn’t caused tremors in keyword rank trackers like SEMrush or Ahrefs. But that’s because most of their datasets rely on short or mid-length keywords. BERT’s power is in the long tail – those natural, sometimes messy top-of-funnel queries.
What It Means for Content
BERT is laser-focused on informational intent. Think of searches like “Which electric cars have the longest range in the UK”. Generic, broad-stroke content won’t cut it anymore.
You need to be specific. You need to be helpful. You need to be human.
That’s what “searching in a more natural way” really means – not typing better, but having the algorithm do the heavy lifting.
Where It’s Headed
So why all this effort?
Because the better Google gets at serving you exactly what you need, the more you trust it. And the more you trust it, the more likely you are to return, ask more, and yes – click the paid links.
Ultimately, Google wants to be your go-to for everything. Not just a search engine, but a digital crystal ball.
And BERT? It’s just one more piece of that puzzle.
This article originally appeared in a modified form on The Drum, December 2019.
