Google started with a very simple promise: “Give us your search query, and we'll match it to keyword stings within web documents in our index”.
Back then, search engine optimization was mostly about identifying those popular search queries and repeating them throughout your web pages for Google to be able to match them when that query is searched.
Obviously, this method resulted in two main problems:
- It ignored the context (of the query and in the document) so searchers were often served irrelevant results
- It offered too many opportunities to manipulate the relevancy signal by adding searchable queries within an irrelevant document
Google understood both of those weaknesses of that old-school method, so they worked hard to change the algorithm at its core, i.e. to teach the machine to understand queries like humans do.
Two most important query processing algorithms: Hummingbird and BERT
There must have been a lot of behind-the-scenes updates, tweaks and experiments but the two Google chose to announce and explain were Hummingbird and BERT.
Yet, it was more than a tweak to the algorithm.
Search Engine Land compares it to changing the car engine of an old car.
Instead of targeting one specific signal (links, content, etc.), Google announced changing the way they treated search queries, i.e. instead of taking a search query verbatim (word by word), Google was trying to take a query in the context.
This is where we learned that Google was using “things” (i.e. entities) to better understand the context and match to a more relevant document.
From now on, Google was using “things instead of strings”, instead of using “keyword strings”, the algorithm was trying to identify concepts and entities within a query to better understand its meaning.In fact, Google’s massive knowledge graph, which had been introduced the previous year, was becoming more important with Hummingbird. ‘Things, not strings,’ is how Google describes the concept. And Google is expanding on this idea, making the company more of an answer engine than a search engine
BERT was a second big change to Google’s query processing algorithm. It was announced 6 years after Hummingbird, in 2019.
BERT is primarily Google’s effort to understand longer, more conversational queries.
To better understand BERT, watch this quick video from Google:
In it Google explains the struggles the machine is facing when it is trying to understand a simple query. For example, if a recipe instruction was to mix the batter with the banana, you wouldn’t naturally think to use the banana as a mixing device:
But what’s obvious to humans isn’t really easy to understand to a machine. BERT is the combination of machine-learning algorithms helping Google understand human language in order to return relevant results.
Simply put, BERT is Google’s query processing algorithm helping Google understand and process queries like human beings would.
Conversational Search as a Whole
With both of those updates, Google could now respond in a more conversational way when it came to searches. In other words, it could search for what best answered a question, instead of just focusing on keywords.
As mobile was becoming more and more important, people were searching more by literally asking questions and speaking to their phones. So Google started answering those questions in a more natural way.
Search for things like “Tell me about Impressionist artists’ and get a plethora of information at your fingertips, especially if you are making the search on a mobile phone.
So what does this mean for your SEO strategy?
On a higher level, this means one important thing: There’s no need to stress about exact-match keywords any more and that is true for all major tasks:
- Brand naming
Brand naming: Don’t invest into keywords
The days when keyword-focused domains gave you a huge organic competitive advantage are not so distant: People are ready to invest huge money into a domain name just because it contains its target keyword.
Focus on branding. Think how easy that domain might be to remember and type into the address bar. Consider if it’s unique and original enough. Namify can help you find a cool domain name for as low as $2, and that domain name is sure to trigger niche associations for Google and your customers to easily classify you as a brand:
Additionally, when building your website, make sure to follow the recent performance and rendering standards.
Content optimisation: explore “things” behind the “strings”
While keyword research is still important for you to understand searching patterns and what people tend to struggle with, the actual optimization process is a bit different these days:
- No need to repeat the same keyword lots of times within your content
- Focus on depth and diversity of your content because that’s what Google will be looking at
Text Optimizer lets you explore “things” behind query strings allowing you to discover concepts and entities that make your page match user intent and better cover the topic:
Text Optimizer also shows popular questions on any topic.
Surveying your audience and social media following is another way to uncover important questions to address in your content.
Setting up a FAQ section on your site to cover niche questions is a good idea. You can also set up a course to address questions in more detail there and attract leads. There are quite a few platforms allowing you to create a course easily.It is also important to create diverse content, including visualizations, videos and even slideshows. Given how diverse Google’s SERPs are, you need to produce different formats of content to ensure your brand’s organic search visibility.
Backlinks: Avoid exact-match anchor text
ow that Google has a much deeper understanding of a natural language and context, forcing exact-match text into your link building strategy can do more harm than good (easier for Google to label you as a backlink spammer).
SE Ranking backlink tool is a great way to analyze your site’s backlink profile and identify if it is natural and diverse enough.
When planning your outreach or linkable assets, anchor text should no longer be part of your strategy.
If you have a website, and want to attract traffic to it and you are interested in performance based SEO, you need to focus on answering important questions, and not simply including keywords that people can search for. This makes web quality more important than ever. The more your website has quality information, and answers to questions that people may be interested in, the better it is for you.
Additionally, coming up with alternative traffic sources (which wouldn’t rely on Google) is still a great idea. Start a newsletter or create a social media strategy - both of these tactics will help you make the most of your organic traffic by giving you ways to bring those visitors back to your site.
Overall, Google’s query processing algorithms make low-quality SEO efforts even less effective and time-wasting. Creating high-quality content is what works, in the long run, regardless of how Google really interprets those search queries.