What Bert means for ecommerce: lessons from Natural Language
If content is king, Google’s natural language update BERT appears auspiciously placed to answer long-tail queries with finesse. By its own estimates, Google implemented around 3,200 search algorithm changes in the last year alone, but BERT (an acronym for the less-than-catchy ‘Bidirectional Encoder Representations from Transformers’) certainly asserts itself as one of the more notable.
In an explanatory blog post, Search VP Pandu Nayak hails the tool as “one of the biggest leaps forward in the history of Search”.
Natural language and nuance
All told, around 10% of future search queries are likely to be impacted by BERT. Of course, as an NLP (Natural Language Processing) tool, this AI also dovetails very neatly with the steady rise in voice assistants and associated voice search queries, which tend by their nature to be longer. While BERT’s underlying mathematics may seem complex, its aim is perfectly simple: to decode the nuance and context of longer queries, in order to provide users with more accurate, better quality results to their searches.
Bert vs Rank Brain
BERT differs from Google’s earlier Rank Brain AI, which draws upon its own experience of previous searches when producing results. Whilst Rank Brain analyses the way in which a user interacts with search results including dwell time and click-through rates, BERT pays closer attention to the natural language in the search query itself. It understands ‘text-cohesion’ (the grammatical and lexical structure of a text which provides meaning) and operates on a bidirectional basis to process and interpret natural human language.
According to one of Google’s own examples, if conducting a search such as ‘Math practice books for adults’, this could actually mean the difference between the helpful (in this case, a maths textbook for grown-ups) and the less relevant (maths content for Grades 6 – 8) when it comes to the answers at the top of the search results.
Implications for SEO
Google’s Bert announcement prompted an initial flurry of uncertainty about what the tool may mean for SEO. Google’s own Danny Sullivan appeared keen to temper expectations when he tweeted: “There’s nothing to optimize for with BERT, nor anything for anyone to be rethinking. The fundamentals of us seeking to reward great content remain unchanged.” In response to further probing about timelines and other languages, he later added: “…BERT doesn’t change the fundamentals of what we’ve long said: write content for users..”
Good news for ecommerce
In short: Under BERT, user-centred, highly relevant content that reflects search intent will flourish. BERT is good news for any ecommerce site with clear, authoritative and relevant content. As Jill Kocher Brown has noted, the tool’s ability to assess the entire query, including prepositions, will provide ecommerce with new opportunities to direct organic search traffic to product detail and product grid pages. Those product, category and guides pages that provide naturally written, succinct and unique information will perform well. Meanwhile, exact matches will lose some of their potency by comparison, and SEO keyword stuffing will be further a relegated to a thing of the past.
To hear more about our Search Optimisation solution, and how high volumes of high performing search optimised content has helped our clients to win a greater market share of cost-effective, organic traffic, please get in touch.
More posts from the blog