Google has announced it rolled out a ‘freshness algorithm’ in February which aims to ensure the featured snippets it displays are timely, fresh and relevant when the query demands it
At the core of ‘search’ is language understanding, and Google constantly developing new ways to better understand searches and provide relevant results, especially in cases where there is useful context that is implied, like whether freshness matters.
What’s the difference?
Some searches are evergreen, such as “What date is Christmas Day?” The answer is definite and not time sensitive or subject to change.
But other times you may be seeking information when timeliness matters, even if the query doesn’t make that immediately obvious. For example, when you search for something like “weather forecast”, it’s likely that you’re looking for current, up-to-date information:
Similarly, “who is the US President?”:
Because the freshness of the content in these cases is imperitive for returning the most relevant, helpful and accurate results, Google prioritises this content in its results
How does Google determine the context of the search query?
Google is becoming increasingly adept at identifying the context and intent behind the words and phrases that people search for
This process is known as latent semantic indexing (LSI). This allows the search engine to determine the relevancy of a page not only to the search query, but also to the context of the search query.
They take into consideration topic themes and associations, relating terms, phrase matches and variants. For instance, search engines are capable of identifying that the English language terms ‘vitamin supplements’ and ‘multivitamins’ are conceptually similar.
The images below shows LSI keywords appearing in Google’s search results: the top in Autocomplete in the search box itself; the image below as ‘related searches’ which appear at the bottom of results pages. Google is clearly recognising that ‘squash’ can relates to both the food and the sport, and attempts to satisfy both types of intent:
RankBrain is a machine-learning artificial intelligence (AI) system developed by Google using LSI keywords to help it interpret queries and process its search results. More specifically, it’s used to interpret the meaning and intent of content compared to the query, ‘learning’ from previous data to inform current decisions and work out the intent of queries it’s never seen before to Google displays relevant answers.
For example, if a user is searching for ‘best cereal bar brands’, RankBrain knows that the user is looking for particular types of information. The user may want a list of multiple cereal bar brands, not just one, and is possibly looking for some sort of evaluation criteria such as reviews on a particular cereal bar brand or product. RankBrain is also able to interpret that the user is looking for brand names and not necessarily cereal bar recipes – this type of query interpretation is something that Google, in particular, is very adept at determining. This also applies to sentence meaning. For example, in the past, if a user was looking for ‘best sugar-free cereal bar brands’, Google may have focused on the word ‘sugar’ and returned results containing a lot mentions of ‘sugar’ within the content, and a few mentions of ‘sugar-free’ – hence the results would be less likely to satisfy the user’s search query.
Pandu Nayak, Google Fellow and Vice President, Search, said: “As part of our ongoing efforts to make Search work better for you, a new algorithm update improves our systems’ understanding of what information remains useful over time and what becomes out-of-date more quickly. This is particularly helpful for featured snippets, a feature in Search that highlights pages that our systems determine are most likely to have the information you’re looking for. For queries where fresh information is important, our systems will try to find the most useful and up-to-date featured snippets.”