top of page

What is the Google algorithm?

an image of author Mordy Oberstein, accompanied by a graphic of a browser. The text on the image reads "What is the Google algorithm?"

It’s impossible for Google to manually sort through billions of web pages in order to match the best ones to a given search term. That’s why Google employs algorithms to provide searchers with a relevant set of results in a scalable manner.


More accurately, Google uses a collection of algorithms that work harmoniously to offer the right kind of information for the given search term. 


That makes understanding what Google’s algorithms are and why they exist vital for every digital marketer, brand, business owner, or person that needs to get traffic to their website from search. 


Let’s get started with what you need to know.


Table of contents:




What is the Google algorithm?


What digital marketers refer to as the “Google Algorithm” is a collection of integrated algorithms and ranking systems that determine which results and features Google shows for a given search term or keyword. 


These systems use data from web crawls, user behavior, machine learning, and other signals to sort and rank web content. So rather than being a static entity, the so-called “Google Algorithm” is made up of many constantly moving parts.


Why SEOs pay attention to Google’s algorithms


Changes to the Google algorithm (also known as “algorithm updates”) can sometimes result in sudden traffic and visibility drops for websites. The professionals supporting search engine optimization (SEO) clients at digital marketing agencies or consultancies keep track of these updates to fine-tune their SEO strategies and approach.


By monitoring and assessing the impact of Google algorithm updates on your website, you can create a plan to address changes in your organic traffic.


Why Google updates its algorithms


Since new content is constantly getting added to the web, the machine learning behind the algorithms constantly recalibrates how it understands content, adds additional considerations to the mix, and makes changes. 


Google also wants to ensure that it meets the changing needs of users and attempts to align its algorithm with evolving consumption trends. When Google makes changes to its ranking systems, big or small, these are called algorithm updates.


Google’s most important algorithms and ranking systems


Though Google has never disclosed a complete list of the criteria it uses for ranking, the company does share some information around some of its most important ranking systems. 


For those working in SEO today, knowledge of the following algorithms can help you better manage your SEO priorities:


  • PageRank

  • Hummingbird & RankBrain

  • BERT

  • MUM

  • Helpful Content System & the Core Algorithm 


PageRank

Google’s original search algorithm is called PageRank and it has been a central part of the company’s algorithm since it was founded. This algorithm heavily relies on the number and quality of backlinks a site/page receives to rank content.


An illustration of the PageRank algorithm.
An illustration of the PageRank algorithm. Source: Wikipedia.

This model views links as an affirmation of a page’s content quality—if high-quality sites linked to a page often, then Google assumed it must contain quality content. 


The problem, however, is that links are a secondary signal. The number of links doesn’t tell the search engine about the actual content quality, only the probability that it contains quality content (since many people linked to it). Websites could easily game the algorithm through spammy links and illicit content practices. 


Hummingbird & RankBrain

In 2013, Google essentially restructured its core algorithm to focus less on linear interpretation and more on semantic understanding. The update, known as Hummingbird, introduced the concept of “things, not strings” to the SEO vernacular. 


In short, Hummingbird gave Google the ability to understand context to better connect various concepts and “things” together. This resulted in the search engine being able to move beyond directly matching the terms used in a search query with the content on a page. 


In this way, Google was able to understand that when someone searched for [New York’s oldest pro hockey team], for example, that the intent was to find the New York Rangers. Moreover, the ranking pages would not need to necessarily include the term “New York’s oldest pro hockey team” to rank, as Hummingbird was able to make the semantic connection.

Then in 2015, RankBrain’s release ushered in the age of machine learning. RankBrain extended and amplified what was possible with Hummingbird (to the best of our understanding, RankBrain operates “within” HummingBird). 


With RankBrain, Google could analyze a massive amount of user data to see what pages satisfied users’ intent when searching for a specific query. With this information, RankBrain could build models to understand which on-page factors or content signals were more (or less) important to a searcher and use them to structure search results.

For example, when you searched for [buy car insurance] prior to RankBrain, the search results page would list car insurance providers. However, Google quickly realized that, for such queries, users often wanted information about buying the product or service (in this case, car insurance). Now, Google shows a healthy mix of providers that educate users on how to choose the best insurance:


ALT A screenshot of the google search results for “buy car insurance” showing a listing for a page titled “how to buy car insurance online for 2022”

Similarly, because of RankBrain, Google discovered that users preferred an image of the finished dish alongside recipe results. To now rank for a recipe-related keyword, a page must contain an image (all things other considerations being equal):


A screenshot of the google search results for “healthy stew recipe for diabetes” showing a carousel of recipe results, each containing an image of the finished dish.

Moving beyond secondary signals, Google has implemented technology like neural matching, Bidirectional Encoder Representation from Transformers (BERT) and Multitask Unified Model (MUM) to better understand user queries.


Diagrams showing how BERT, GPT, and ELMo work.
Source: Google.

BERT

Google’s BERT (Bidirectional Encoder Representation from Transformers) is a machine learning algorithm that helps Google codify the relationships between words and word sequences to better interpret the meaning of content. This is considered to be a foundational algorithm and is used to identify sentiment, questions, and predict text with more natural language. 


Formally introduced to Google Search in 2019, Pandu Nayak, the company’s VP of search, explained that BERT models can “consider the full context of a word by looking at the words that come before and after it—particularly useful for understanding the intent behind search queries.”

For SEOs, this change signals the introduction of more emphasis on semantic search, the intent behind keywords, and also the implicit intent of a given search.



example of BERT impacts search
Source: Google.

MUM

Publicly launched in 2021, Google’s Multitask Unified Model (MUM) has 1000x more machine learning power than BERT and signaled an intention from Google to better integrate multimedia into its search queries and user journeys


The company teased the potential applications of this new algorithm via significant integrations with Google Lens and visual search as ways to search Google. Like BERT before it, this algorithm is designed to help Google better understand content, including multimedia, rather than to assess the quality of a website.


A gif depicting how MUM helps Google interpret a query and offer relevant responses.
Source: Google.

To date, MUM’s application has been limited compared to other algorithms and ranking systems. MUM was first implemented to better help Google understand the various ways searchers were referencing COVID vaccines across the globe. MUM also helps Google improve quality by identifying consensus on the web. As time goes on, the most logical assumption is that Google will find additional ways for MUM to play a role in the ranking process. 


Helpful Content & the Core Algorithm 

In August 2022, Google released the Helpful Content Update (HCU), an algorithm designed to target low-quality “that seems to have been primarily created for ranking well in search engines rather than to help or inform people.”


The HCU follows in the tradition of historic, now-retired algorithms, like Panda, which tended to reward (and devalue) domains based on perceived quality. But unlike those updates, the Helpful Content classifier “runs continuously, allowing it to monitor newly-launched sites and existing ones,” Google explained, “As it determines that the unhelpful content has not returned in the long-term, the classification will no longer apply.”


At Pubcon in February 2023, Gary Illyes, analyst on the Google Search team, suggested that recovering organic visibility after being negatively impacted by the HCU may be particularly difficult. And although SEOs observed minimal impact from the HCU’s initial release, following updates in 2023 showed this algorithm’s significance. 


In March 2024, Google announced it would no longer rely on one system to determine “helpfulness” but rather would assess sites for “helpfulness in a multitude of ways, following the path of previous heavy-weight algorithms such as the aforementioned Panda update as well as Penguin (which focuses on devaluing spammy link practices).


Having the core algorithm determine helpfulness prevents a scenario where the HCU competes with other algorithms. Think of the “core algorithm” as a stew, where each spice and ingredient works in relative harmony with the others. An update to the core algorithm might mean a change in how those various ingredients factor into each other and the role they play in the overall stew—among other things (such as advancements that enable the elements within the core algorithm to function at a higher level).


Do ranking factors affect Google’s algorithms?


Don’t get confused: While the SEO industry has long discussed specific “ranking factors” or “ranking signals,” there’s a lot of misinformation out there and I strongly advocate for a holistic approach.


Google allegedly uses more than 200 official ranking signals to decide what content should (or should not) rank for a query. These factors include anything from links to content relevance. Many SEO professionals debate one ranking factor’s importance over another, while others will try to “optimize” for as many “factors” as possible.


This is a mistake.


For starters, there is no universal list of the most weighted factors. The factors Google uses vary from query to query and from vertical to vertical. Google uses a complex process to evaluate content, employing machine learning to understand language and better classify and profile content to determine its quality:


“It’s something where, if you have an overview of the whole web (or kind of a large part of the web) and you see which type of content is reasonable for which individual topics, then that is something where you could potentially infer from that. Like for this particular topic, we need to cover these subtopics, we need to have this information, we need to add more images or fewer images on a page. That is something that perhaps you can look at something like that. I am sure our algorithms are quite a bit more complicated than that, though.” — John Mueller, Senior Search Analyst at Google via Search Engine Roundtable

For example, trustworthy medical content can greatly impact a reader’s health, so it should look, sound, and feel different than a gossip column. So for health information, Google trains a machine learning model to identify a content profile based on its leaders, such as the Mayo Clinic.


The assessment of “good” vs. “bad” content happens in what I’ll call the “pre-algorithmic” or “meta-algorithmic” stage and impacts what kinds of pages rank beyond a specific ranking factor.


Simply put, don’t worry about “ranking factors.” Instead, write highly-targeted, substantial content. Cater your content to your target audience’s needs, knowledge level, and frame of reference. Make sure your content genuinely helps them. That’s the most important thing.

Google aligns its algorithms to users


Though the tooling may change, ultimately each Google algorithm works to more closely understand user behavior, to put users first. So to better understand Google’s algorithms, it helps to look at your site and its pages from the perspective of a new user. 


What signals are you latently sending? If your tone is not serious enough (or perhaps too serious given the niche), what will users feel? 


When you take a step back and look at the subtle signals your pages give off, you are aligning with Google. You’re putting yourself in a position to go beyond Google’s current capabilities, which, in turn, puts your site in a prime position to improve its rankings as algorithms change or when a core update releases.


 

mordy oberstein

Mordy is the Head of SEO Branding at Wix. Concurrently he also serves as a communications advisor for Semrush. Dedicated to SEO education, Mordy is one of the organizers of SEOchat and a popular industry author and speaker. Twitter | Linkedin




Get more SEO insights right to your inbox

* By submitting this form, you agree to the Wix Terms of Use and acknowledge that Wix will treat your data in accordance with Wix's Privacy Policy

bottom of page