3 Ways to Optimize for Google's BERT

4.4
125
3 Ways to Optimize for Google's BERT

Google recently announced that they would be using BERT in search results, which means Google Search will rely on their new generation of neural networks.

Google recently announced that they would be using BERT in search results, which means Google Search will rely on their new generation of neural networks to understand written content more effectively. While there are still many questions surrounding BERT's effect on search results, marketers will likely have to optimize their websites and content to increase the chances of being ranked higher in Google Search as a result of BERT's influence. In this article, we'll give you three ways to optimize your website or content so that Google can better understand it based on their new BERT technology.

1) Use visual content

For better or worse, Google is becoming more image-centric. (Worse if you're an SEO or designer; better if you want people to understand what your brand is all about.) As part of its emphasis on visually related words, Google announced it would be using deep neural networks (aka neural reordering techniques based on Recurrent Neural Networks, aka Google's Word2Vec model, written in TensorFlow and open-sourced last year.

If you speak brainiac fluently—or are willing to do a bit of homework—then there are three ways you can start optimizing your content, so it supports Google's latest AI-powered approach.

2) Write a long-form copy

Research suggests that long-form copy can have a positive impact on engagement. When it comes to optimizing your site and copy, long-form content tends to drive more organic search traffic than short snippets. Since clickthrough rates (CTRs) are so low across most websites anyway, long-forms could be an area where you see some strong improvements. By default, Google also favors longer pages in its SERPs; anything between 800 and 1,500 words is considered ideal by Google's Search Quality Evaluators.

3) Use Rich Snippets

BERT stands for Bidirectional Encoder Representations from Transformers. It is essentially a representation of language that allows machines to translate language without requiring pre-processing. The original model requires some pre-processing of language, while Google's newer model—BERT—requires no processing. A good way to use BERT in your SEO strategy is through Rich Snippets, which allow you to attach additional information on search engine results pages (SERPs).

To do so, create one-sentence summary descriptions with appropriate keywords and tags so users can quickly find what they are looking for. Rich snippets are updated directly by Google Search rather than on your site or CMS, so there is little effort required on your part.