What is the Google Knowledge Graph?
A knowledge graph is a collection of entities (people, places, and things). Each entity is represented and referred to as a node on the knowledge graph. The knowledge graph also includes relationships between those entities, known as edges or tuples.
Tuples (triples) consist of a subject-predicate-object (SPO). An example of a triple is “Mary owns Mary’s Hair Salon.” With Mary being the subject, owns being the predicate, and Mary’s Hair Salon as the object. These connections on a knowledge graph help computers understand relationships between entities and more closely resemble how people think. Knowledge graphs can also be thought of as knowledge bases since they can include billions of entities and can contain trillions of facts about those entities.
Google’s knowledge graph was launched in the U.S. in 2012. With this update, Google now has the ability to understand things (entities) rather than just strings (keyword matching). Google’s knowledge graph has grown quite rapidly and as of the last official update in 2016, the knowledge graph consists of 5 billion entities and over 500 billion facts about those entities.
Semantic SEO has become more important than ever as Google is now able to understand the true meaning behind your content. Search engines are better able to understand the breadth, depth, and value of content for users on a deeper level. Optimizing your website and entities for semantic search and the knowledge graph has become an important focus of any SEO or marketing strategy.
In a recent Twitter conversation with Bill Slawski (SEO/Google patent researcher) and Jason Barnard (Brand SERP Guy), Jason took an attempt at estimating the number of entities that are currently on the knowledge graph.
Jason estimated that there are currently roughly 9.25 billion entities on the knowledge graph with over 925 billion facts about those entities.
When do you see the knowledge graph in search?
The knowledge graph originally appeared on knowledge panels or knowledge cards. These results are commonly triggered when a user searches for a query that includes well-known entities like organizations, notable people, and more. Another few areas that are powered by the knowledge graph that we will touch on include voice search and featured snippets.
Knowledge Panels
Typing in the query “Nike” should result in a knowledge panel.
Looking at the knowledge panel results, you can see a large amount of information. Founding dates, contact information, executive information, social profiles, competitors, and much more are all sourced into the knowledge graph from multiple sources. In order for a particular result to end up on the knowledge panel, Google must be able to corroborate that the information is a fact (or as close to a fact as possible). This is an important aspect of the Google Knowledge Graph. To ensure the knowledge graph is accurate and serves factual information to users is the highest priority. With this in mind, Google will typically only add entities and facts to the knowledge graph that it is able to validate across multiple reputable sources (more on that later).
Knowledge Cards
Knowledge cards appear a bit differently in SERPs. Knowledge cards are commonly triggered when users search for a particular fact about a known entity.
So searching for the query “when was Nike founded?” should return a knowledge card result.
We can see that Google search results returned the date and location that Nike was founded, as well as some competitor information. This is Google’s attempt to provide the best user experience possible.
Featured snippets
The knowledge graph is also responsible for search results that show a featured snippet. These types of results feature very prominently in search and typically pop up when asking a question. These types of results can have high click-through rates, raise brand awareness, and more.
Featured snippets have been responsible for a drop in organic search clicks, but winning featured snippets for a business can be a game-changer.
Optimizing your questions to answer the query immediately—or at most within the first few sentences— is one crucial tip to earning featured snippets.
Voice Search
The knowledge graph is also responsible for Google voice search on devices such as Google Home and Google Assistant.
What is a no-click search?
At this point, I would be misleading you if I didn’t mention the impact the knowledge graph had on no-click search queries.
In December 2020, SparkToro ran a study to determine the number of queries that end in a click vs no click and the results were astounding. Nearly 65% of search queries ended with no click. An increase of over 15% from a previous Search Engine Journal study conducted in 2019. This growth of no clicks searches directly correlates with the launch of the knowledge graph in 2012.
You can see that more and more searches are resulting in no-click. This makes it more important than ever to ensure you are optimizing your brand for these no-click results. If you need a brand expert that you can trust, feel free to call the search engine experts at Webfor.
What are the benefits of Google’s knowledge graph?
Google’s knowledge graph has brought many benefits to both searchers and the SEO community. The knowledge graph and semantic SEO have brought about a new experience to the SERPs.
- Better search experience
The knowledge graph is able to provide answers to users real-time in SERPs. The knowledge graph enables Google to extract entities or even rewrite queries with entities to better answer a user’s query. This results in fewer clicks and a better user experience. - Better understanding of topics
The knowledge graph has taken Google’s understanding to the next level. It has transformed it from what resembles keyword matching, to something more like how you or I would understand relationships between entities. - Higher CTR
Another benefit of the knowledge graph is the high click through rates that can be received from earning featured snippets. These prominently positioned rich results tend to receive a large number of clicks due to the visibility and amount of real estate they take up at the top of SERPs.
How does the knowledge graph work?
As mentioned previously, Google’s knowledge graph has grown exponentially. With the last official Google reports showing 5 billion entities, one may wonder where Google built this knowledge graph of entities and relationships.
Google builds the knowledge graph from multiple sources. The initial primary sources of data included:
- Wikipedia.org
- CIA Factbook
- Firebase
These are not the only sources Google uses to build its knowledge graph of entities. Here are some other sources that Google is using to source its knowledge base.
-
Reputable websites
It may not surprise you that Google sources this data from the web. One of the primary sources of entities and facts comes from reputable (high authority) websites. Some of the top websites that Google sources this data from includes:
- CrunchBase
- Bloomberg
- Reuters
- BusinessWire
- Schema
Another area that we have not discussed yet that plays an integral role in Google’s knowledge base, is schema.org. It is one of the main triggers an SEO can pull to help Google better understand its brand, organization, and content. Schema.org is a project that was started by search engines Google, Yahoo, and Yandex on June 2, 2011. Structured data was created to be machine-readable code that can be read across all search engines, which helps search engines better understand the content, intent, topical relevance, content vectors, and more. If you would like to dive deep into what schema markup is check out my previous post on schema markup. Some of the top types of schema that can help Google better understand your entity include:
- Organization
- LocalBusiness
- FAQ
- Products
These are just a few of the important structured data types that need to be marked up throughout your website to help Google.
After sourcing and validating entities from the above sources, Google is then able to add entities and facts to the knowledge graph after they are able to validate the trustworthiness of the information. This is one reason Google’s knowledge graph trusts Wikipedia.org is because of the level of accuracy of information.
Google typically has to validate or corroborate this information across multiple reputable sources in order to add information to the knowledge graph. It is absolutely key for Google to confirm the data through these corroborations, and as Jason Barnard (Brand SERP Guy) shares 30 corroborations as the number of confirmations, Google needs to validate information for the knowledge graph.
Why does the knowledge graph matter for SEO?
In the days before the knowledge graph, users would have to go result by result until they found the content that answered their question. Search engines were only able to understand basic keyword matching. The knowledge graph took Google’s understanding of search intent to the next level. Google was now able to infer intent and differentiate between entities. Semantic search to the rescue!
When typing in a question about who won the 1985 Super Bowl, we can see a knowledge card and knowledge graph.
Taking this a step further by typing in “basketball player with 3 point NBA record,” we can see that Google is able to understand that we are talking about Stephen Curry. Additionally, it provides a knowledge card and knowledge graph with the number or threes he has made, connected entities, and more. These types of results were not possible prior to the knowledge graph.
This has taken user experience to the next level, and continuing advancements to the knowledge graph will continue to improve semantic search for users and search engines.
In addition, it’s possible that having an entity on the knowledge graph could even provide ranking benefits. Having an established entity could help reinforce expertise, authoritativeness, and trust (E.A.T.) with Google. Building out your topical authority on highly relevant topics should be at the core of any SEO strategy and semantic search principles. Having a node on the knowledge graph means Google has a stronger understanding of your brand. This could directly impact how authoritative and relevant Google feels your website and content may be.
7 best practices to get your business on the knowledge graph
1. Wikipedia
Establishing a Wikipedia page will help your entity get in the knowledge graph. Wikipedia is a commonly cited source for the knowledge graph and is one of the sources that Google relies on to build out its semantic web entities (people, places, things) and facts.
Keep in mind: Building out a Wikipedia page can be quite challenging as they have a high standard of notability that must be reached. There are also conflicts of interest (COIs) that need to be considered. This means you cannot build out a Wikipedia page for yourself or your business. This would be deemed a conflict of interest and more than likely will get flagged by moderators and Wikipedia and your page can get removed.
One advanced level tip is to look out for red links. These are links to content/topics that aren’t written about and do not currently exist on Wikipedia.
Red links like the above that mention you or your brand could be helpful to getting your Wikipedia content published and approved.
2. Wikidata.org
Wikidata.org is a multilanguage knowledge graph owned by the Wikimedia Foundation. All Wikimedia projects such as commons and Wikipedia use Wikidata.org as a centralized linked data repository.
Getting a wikidata.org page established and approved is slightly easier than Wikipedia.org. The levels of notability appear to not be quite as strict. But a wikidata.org page needs to be built in a way that shows relevance and relationships between other known entities.
Another note of relevance for Wikidata.org. After the shutdown of Firebase in 2014, Google migrated all of Firebase’s data to Wikidata.org.
When you are completing your Wikidata.org page make sure to fill out as much information as possible, including things like CEO, adding images, industry, and more.
3. Schema markup (organization, person, product, HowTo, etc.)
Schema markup is a structured data vocabulary that helps search engines better understand the content and entities that are represented on your website.
Schema is a powerful tool in an SEO toolbelt that can be used to mark up entities including people, places, things, and facts. Here is a quote from Google on how they use schema.
“Google uses structured data that it finds on the web to understand the content of the page, as well as to gather information about the web and the world in general.”
Google Search Central
This better understanding and gathering of information can result in rich results in SERPs, ranking content in SERPs for more relevant search terms, and more benefits.
To learn more about what is schema markup and how to use it to grow your website’s visibility check out my recent post on schema markup.
4. On-page optimization (about page – connect to known entities)
Your about page should be an important part of your SEO strategy for optimizing your entity for the knowledge graph. The about page on a website should provide detailed information regarding facts about your business. Here are a few examples of important content that should be included on the about page of your website.
- Founding date
- Founders
- Related well known entities
- Industry groups and partnerships
- Charities and other sponsorships
- Industry awards
On top of this, it is highly recommended to use AboutPage schema markup on your about page to help highlight these entities for Google.
5. Google Business Profile listing + optimization
It shouldn’t surprise you that Google Business Profile – previously Google My Business – is another way to help Google better understand your business, and earn your business a spot on the knowledge graph.
This helps Google validate information such as your name, address, phone number (N.A.P.), business categories, founding date, and much more. Google uses this information to corroborate the validity of your entity that it is finding throughout the web.
If you haven’t set up a GBP profile for your business, head over to business.google.com and click “get started” to start creating your GBP profile.
6. Entities (Knowledge graph search API – JSON-LD/Schema markup types)
This isn’t necessarily a technique by itself that will help you get your entity in the knowledge graph, but it’s important to keep in mind when working on building out your knowledge graph.
Marking up well-known entities that are already established on the knowledge graph and then connecting those entities to your business can help Google “connect the dots.” As a result, this increases your likelihood to earn a spot on the knowledge graph. For example, if you are a well-known technology business that wants to get yourself in the knowledge graph, you can use your connections to well-known entities to help Google build out your node on the knowledge graph.
Use schema to mark up industry groups you are part of, non-profits you partner with, and more.
Another advanced technique is to use the Google Knowledge Graph Search API to analyze entities that are on the knowledge graph. This helps you understand what types of content Google includes in the knowledge graph so you can ensure you are doing all you can to highlight and provide this information for Google.
7. High authority sites (Crunchbase, industry-relevant authority sites, social media, etc.)
That’s right, Wikipedia.org is not the only website that is valuable for your brand to earn a spot on the knowledge graph. Other websites that are valuable for your brand include:
1. Crunchbase
2. LinkedIn.com
3. Bloomberg.com
4. Reuters.com
5. Facebook.com
6. Businesswire.com
7. Twitter
These are just a few of the high authority sites that can help corroborate your business for the knowledge graph. Other examples of high-value sites include industry-relevant sites, additional social media sites, and more.
How to suggest changes to your knowledge panel?
Looking at Arnold Schwarzenegger’s knowledge panel above, at the bottom we can see the button with the anchor text “Claim this knowledge panel.”
After you click the button to claim the knowledge panel, you will then be directed through a verification process to claim your listing. Depending on your entity, Google could require verification through a picture of your photo ID, Google Search Console access, or another source of verification.
In the above image, you can see the success message you receive after you claim your knowledge panel. Once you complete the verification process you will then be granted access to suggest edits to your Google knowledge panel.
If you are having trouble updating your knowledge graph, refer to Google Knowledge Panel Help – Update your knowledge panel documentation for assistance.
What’s going on with the knowledge graph now?
Rewritten queries and user-specific knowledge graphs
Google’s patent for on-device query rewriting shows how Google can use a user-specific knowledge graph to anonymize and rewrite queries that are specific to facts it knows about you, your spouse, friends, or others. For example, if you type in “where should bob and I go to dinner?” Google is able to rewrite this query to anonymize the data and assign and entity and a category that can help improve search results.
Like in the above query examples Google can use information from Google calendar invites, emails, and the user-specific knowledge graph to serve search results.
Following the above examples, after searching “where should bob and I go to dinner?” Google can then use the user-specific knowledge graph to understand what results are most relevant for you and Bob. So if your user-specific knowledge graph on your device knows that you dislike Italian food, it can provide relevant results that exclude Italian food restaurants based on the knowledge attained through your user-specific knowledge graph. This may be done by rewriting a search query with a category and an entity to help improve results returned to a searcher.
In the above picture from the same Google patent on device query rewriting, we see how Google may rewrite queries to improve results. Google may identify references to a known entity on the knowledge graph and a category.
SEO and Google Patent researcher Bill Slawski’s article, Rewritten Queries and User-Specific Knowledge Graphs covers the above subject in greater detail.
Knowledge panel autocomplete predictions
We are now seeing Google pull knowledge panel results directly into Google autocomplete predictions. This provides a more interactive and better experience at time of search. Google appears to be providing relevant results based on queries that are directly related to the entities you are searching for.
In the above example, we can see people also searched results and related questions being pulled in at time of search in addition to the typical autocomplete predictions.
Conclusion
Semantic SEO continues to grow and should be a focal point of an SEO strategy. Optimizing your brand for the knowledge graph is crucial to building the topical authority and relevance that Google is looking for.
The above article scratches the surface on the Google knowledge graph and why it is important for SEO. We also dug into some tips for getting your entity added to the Google knowledge graph. Check out my recent blog post and explore how to analyze entities using Python. We then looked at what’s new on the knowledge graph and how it is becoming more powerful. If you want the SEO experts at Webfor to help analyze how you are approaching semantic SEO, check out our SEO services to get started.
You can follow me on Twitter @ MPCTheMarketer or on LinkedIn at Michael Patrick Cortez to keep the conversation going on social. Have any thoughts on the knowledge panel and Semantic SEO? Drop a comment below and let’s connect.
