Google’s BERT natural language search technology has been with us now for several months, since late Oct. 2019. The question for marketers is, have you explored what difference BERT is making in terms of how Google treats your website with respect to the keywords you most care about? In this post, we’ll look at some steps you can take to make sure you’re taking advantage of opportunities to raise your search profile in the age of BERT.
Background on Google BERT
First, a bit of background. BERT, which is short for Bidirectional Encoder Representations from Transformers, uses neural network technology to perform natural language processing (NLP). In simple terms, it’s technology that helps Google better understand the intent behind the terms users employ in searches.
An example from a Google blog post illustrates nicely how BERT works. Prior to BERT, the search phrase “2019 brazil traveler to usa need a visa” turned up information about U.S. citizens traveling to Brazil, whereas the clear intent of the query is the other way around – for Brazilians traveling to the U.S. The preposition “to” is crucial to the query – and BERT gets that. Post-BERT, for the same query Google provides more relevant results.
BERT’s reliance on natural language processing raises the question of whether keywords still matter, or how much they matter. After all, if BERT can “read” text and understand context, what’s the point of keywords? (A pair of Microsoft and Google execs had an interesting back and forth on that topic here.)
Our take is in the BERT era keywords certainly do still matter – but the context surrounding them matters more so than in the past. Because it’s in that context where “intent” lies.
The good news is Google gives you tools that can help you understand how your site is ranking for keywords you care about as well as hints for how to get better – namely, Google Search Console and Google Analytics. (Pro tip: Set up Google Analytics to pull in data from Google Search Console. It’s simple to do and will make it far easier to analyze results.)
A first step is to find keywords you already rank for that get a decent amount of search volume. Look for terms leading to pages that are ranking on page 2 or perhaps 3 of Google, what’s known as “striking distance” of page 1. Then the idea is to optimize the content on the pages that are ranking to try to boost the pages up the search ladder. That’s where the natural language processing part comes in, by creating content that uses the appropriate terms in natural, meaningful context. Google likes that – now more than ever.
Here’s an example from our own website, SaratogaB2B.com, using Google Search Console. A term we care about is “B2B SEO agency” because it describes what we do and, according to the Google Ads platform, has good monthly search volume. Prior to BERT, our average search results position for the term was 62 – not at all within “striking distance” of page 1 results. But during the month of Jan. 2020, our average position was 17, on page 2 of results. For the term “B2B SEO agencies,” a similar thing happened: we went from an average position of 83 to 21.5. And for terms with even more obvious user intent (but lower search volume), such as “looking for a b2b SEO Agency,” again, we’re now on page 2 for the key page that describes our B2B SEO services (See Figure 1.)
After the official debut of Google’s BERT natural language processing engine in October 2019, SaratogaB2B.com’s average position for search terms related to “B2B SEO agency” shot up. The orange lines show pre-BERT positions while the green is after.
What accounts for the change is we have quite a bit of content on the site describing our B2B SEO services and technology. With the advent of BERT, and perhaps thanks to a few optimizations we made, it appears we’re earning more credit in Google’s eyes for our content than we were before. In that respect, BERT can be a godsend if you’ve got good, in-depth content on your site about topics your customers care about. If not, put strategic content development on your to-do list for Q1 and beyond.
Among the small steps we took in the last four months regarding content was changing the title tag and some verbiage on the main “B2B SEO services” page. With BERT, even small, targeted tweaks like that can better inform Google about the intent of the page, to your benefit.
Quality content is key to success with BERT
Content development, by the way, should be an ongoing endeavor, given BERT will be evolving and getting sharper over time. The more you continually add to your key pages, making your “intent” story stronger, the more you’re likely to impress Google. A monthly or quarterly review of the changes in your Google performance would be a terrific idea.
So, the takeaway here is to look at Google Search Console to find out where you’ve got a decent amount of search activity (impressions or clicks) around search terms that are relevant to you. (Use Google Analytics, too, if you want to see how your landing pages are faring.) Be sure to look for clusters of terms that are semantically related and focus your content development and optimization efforts around them. At the same time, look for areas where you are deficient in performance, and identify content gaps you need to fill. If you can write good, in-depth content that uses the terms in a natural way, Google’s BERT may reward you handsomely.