By Christian Stobitzer, Senior SEO Manager at IONOS
We’re surrounded by them every day. We touch them, interact with them and even Ancient Greek philosophers spoke about them – albeit in a rather different context. I’m talking about entities.
The term comes from the Late Latin word “ens”, which means “being” or “thing”. Despite being ubiquitous, entities are mainly found in philosophical contexts.
However, they also appear in the context of IT, such as in an entity-relationship (ER) model. Due to the rise of natural language processing, entities are also becoming increasingly important for other disciplines.
One area set to be affected by entities more than any other in the years ahead is online marketing. Yet even today, entities have declared war on keywords. But why is that – and how can you optimise your own entity? Read on to find out.
What are entities?
The term “entity” is quite difficult to define – in fact, it can vary significantly depending on the context. In this context, that of online marketing, we’ll be drawing on the definition Krisztian Balog proposes in his book Entity-Oriented Search. He describes an entity as follows:
“An entity is a uniquely identifiable object or thing, characterised by its name(s), type(s), attributes, and relationships to other entities.”
This might all sound abstract at first, but we can illustrate it very easily with an example.
Example: Boris Johnson is an entity
The UK Prime Minister, Boris Johnson, is an entity – just like you and me. Boris Johnson is characterised by his name, but that doesn’t allow us to identify him conclusively. Ultimately, there could well be other people in the UK who are also called Boris Johnson. This is why entities are also characterised by their types.
Types are effectively classifications that “group” multiple entities together. One entity can have several types. A very general type for Boris Johnson, would be “person” – and, as the subtype of that, “man”. In the case of Boris Johnson, a far more specific type would be “Prime Minister of the United Kingdom”.
This means we can identify Boris Johnson as a unique entity based on his name and one of his types. Not everyone is Prime Minister of the United Kingdom – but it is still important we can identify individuals conclusively, which is why we can also use entities’ attributes. For Boris Johnson, for example, this would include his place of birth, his birthday, his age, his height, his salary, and so on.
Another very important aspect for entities is their relationships with other entities, which allows us to draw a lot of conclusions about them. For instance, if all we did was examine Boris Johnson’s relationships with other heads of state, even if we didn’t know who he was, we’d be able to infer that he must hold an important position in the UK.
Relationships between entities mean that Google is now in a position to accurately answer questions such as “Who is Boris Johnson’s girlfriend?”. This search query perfectly demonstrates why entities will continue to grow in importance for search engine optimisation and online marketing in general in future.
How entities are changing search behaviour
Previously, search queries were generally made up of a single word or a combination of two or three words at most. Human language shrivelled to a fraction of its true depth and complexity when using a search engine. This was very much a necessity: the more words a person packed into a search query, the less accurate the results returned by the search engine. In effect, this trained search engine users to search the near-infinite diversity of online information with just a few words – which didn’t always deliver the best results.
But times are changing.
Constant improvements in automated natural language processing and the emergence of new input devices, such as digital assistants, have led to search queries becoming increasingly long and complex. However, most of these search queries relate to an entity. They might be named in the search query, either directly – “How old is Boris Johnson?” – or indirectly – “Who is the Prime Minister?”. In both cases, the search query concerns the entity Boris Johnson. As a result, it is increasingly important that search engines can identify entities and websites related to them. This is particularly true for complex search queries.
Conventional keyword searching is becoming increasingly irrelevant, as is SEO based solely on keywords. This is particularly due to the fact that new input methods make it possible to make search queries without using words. Imagine the following situation: You’re looking at a photo and want to know who’s in it. The classic approach of making a search query using words describing the photo is unlikely to be successful – as the example below illustrates:
Keywords are, in truth, just one way of describing an entity. It’s commonly accepted that words aren’t always enough to capture something. Drawing again on the definition of an entity given above, an entity is characterised among other aspects by its attributes. These attributes don’t necessarily need to be described in order to exist; depending on the entity, we might be able to feel, smell, hear or just see them.
This is where new input methods come into play. The Google Lens app, for example, enables Google to “see” attributes. If you open the app on your phone and point it at an entity (such as a person or a famous building), it will be able to recognise them:
What’s more, the smart Google Assistant is even capable of listening; it only needs to hear a few seconds of music to determine what song is playing and which artist recorded it:
Keywords need to be rethought in future, taking these considerations into account. Google, Bing and co. are already facing up to the problems raised by the inherent complexity of language. Just using a keyword such as “beetle” doesn’t include any information about the entity – context is needed to determine whether the entity in question is the insect or the iconic car.
However, different keywords can also be difficult to understand without context. Let’s take the London slang term “creps”, for example, which is a relatively new slang term for the far more common term “trainers”. The term is primarily used and understood by young people. However, if a search engine can identify that “creps” denotes an entity with the same types, attributes and relationships as “trainers”, then it soon becomes clear that the two words must in fact refer to the same entity.
Google’s Knowledge Graph
Search engine Google identified early on that looking solely at keywords would become increasingly limiting over the long term. That’s why, in 2012, the company started the so-called Google Knowledge Graph under the motto “things not strings”. Google’s Senior Vice President Engineering at the time, Amit Singhal, described the Knowledge Graph as follows:
“This is a critical first step towards building the next generation of search, which taps into the collective intelligence of the web and understands the world a bit more like people do.”
Google uses the Knowledge Graph to record entities and map their relationships. Nodes and branches are common graphical elements. In the Knowledge Graph, a node represents the entities or entity types along with their respective characteristics, while the branches depict the relationships between these nodes.
The image below demonstrates how the Google Knowledge Graph works. On the left-hand side of the image, the entity George Washington is described with what is known as a “knowledge panel” beside the search results. The right-hand side illustrates how the entity might be depicted in the Knowledge Graph. The nodes are shown as circles and the arrows denote branches:
Other Google services can access the information recorded in the Knowledge Graph at any time and process it as needed. For example, Google Lens might also draw information from the Knowledge Graph in the previous example to identify that the person in camera focus is Albert Einstein. However, other Google services in addition to Google Web Search and Google Lens can also access the Knowledge Graph, including Google Image Search and Google Maps.
So, if you want to make an entity visible in a Google service, you need to make sure that the entity is recorded in the Google Knowledge Graph.
Where does Google get its information about entities?
The initial basis for the Knowledge Graph was Freebase, a knowledge database Google acquired in 2010 when it bought a start-up called Metaweb. However, this was discarded in December 2014 in favour of the Wikimedia Foundation’s Wikidata project. Data from Freebase was incorporated into Wikidata, making Wikidata one of the most important data sources for the Knowledge Graph.
As well as Wikidata, though, the Knowledge Graph also draws on a range of sources that Google can choose to extend or limit at any time. In addition to Freebase, when it announced the Knowledge Graph, Google named Wikipedia and the CIA World Factbook as other sources.
Generally speaking, any website demonstrating authority, trustworthiness and a certain level of structure is a potential source. For example, Google collects event data for its Knowledge Graph from event platforms like Eventful:
Beyond this, Google also enriches the Knowledge Graph with information from its own data sources, such as Google My Business. Structured data, such as data encoded with JSON-LD by webmasters, can be used to further expand the Knowledge Graph.
How can I find out which entities Google has recorded?
The data in the Knowledge Graph is not confidential – Google grants users access (to view the data, at least). The Knowledge Graph API allows anyone with a little programming know-how to retrieve data in the Knowledge Graph.
If you want to get an initial impression, you can use the API Explorer to access the Knowledge Graph directly in your browser.
Each entity recorded in the Knowledge Graph is designated a unique ID. In the example shown above, “1&1 IONOS” is has the ID “/m/02rjt47”. Historically, IDs have started with either “/m/” or “/g/”. Entries that start with “/m/” are data carried over from Freebase. Entities with an ID started with “/g/” have been added by Google. ProfitBricks GmbH, a relatively young enterprise acquired by United Internet in 2017, is thus recorded under the ID “/g/11dxpyy3hd”.
The Knowledge Graph contains an almost countless volume of entities. Even back in May 2012, Google claimed to have a record of more than 500 million. These have in turn been assigned different types. Google lists the following 21 explicit entity types:
If an entity falls into one of these types, there’s a good chance that it can be recorded in the Knowledge Graph. In the next section, we’ll look at local business entities in detail and also explore why entities are, to some extent, already more important than keywords in this sector.
A deep dive on local business entities
Local businesses are a particularly interesting topic for entities. This is because people search for them not only using desktop computers but increasingly also with mobile devices. ‘Near Me’ searches play an important role in this context. These are search queries users make to find products, services or shops in their direct vicinity. All the more important, then, to optimise for ‘Near Me’ searches.
In the following explanation, we’ll focus on restaurants as an example of local business entities – but the statements can be transferred to any local business.
At first glance, local business entities appear very similar to other entities. However, there are key differences in certain respects. For example, if a user makes a direct search for a local business entity, Google displays a well-known element on the right-hand side called the knowledge panel.
The search results for DeliBurgers, a burger joint in Karlsruhe, Germany, look like this:
These search results demonstrate perfectly how Google draws on information from the Knowledge Graph to enrich its search results with further useful information. Users can see at a glance the restaurant’s telephone number, opening hours and address, along with reviews.
Google is now able to display this stored information for more specific search queries. Let’s say I’m looking for a burger joint close to where I am. If I search for “burgers near me” on Google.com, the results look like this:
Taking my current location and the stored addresses of all burger joints in Karlsruhe, it would be very easy for Google to provide clear and relevant information. And it does that without adding any sort of location keyword, such as “Burger King Karlsruhe”.
Until now, in some sectors where location was not a decisive factor for service delivery, companies would often work hard to optimise their presence for location keywords. For example, agencies and freelancers would use keywords such as “web design Karlsruhe” to address customers in Karlsruhe – even though their business was actually based in Berlin. The dawn of entity-based searching is changing this. Either an entity is near to the searcher, or it isn’t. Website operators are no longer able to get themselves listed in search queries like this through keyword optimisation alone.
An address is only one attribute of a local business entity which Google can use to limit the results of a search. If, for example, I want to find the best burger joints near me, I can expand my original search query to “best burgers near me” – which suddenly gives me a different search result:
Burger King, which has an average rating of 3.6, gives way to American Diner, which has a far superior rating of 4.5. When search queries include the word “best”, Google automatically applies a “4.0+ rating” filter that only displays results with an average rating of 4.0 or higher.
At the moment, this only works on Google’s English-language variants (such as .co.uk and .com) and not on foreign language versions of the search engine – but it’s safe to assume that this functionality will be extended pretty soon. Once again, though, there’s no way to optimise the position of a business with keywords like “best burger joint”, “great burger joint”, and so on – in future, the naked figures in ratings will decide whether a business shows up in a search or not.
So, to remain visible to search engine users in future, it’s essential that businesses give thought to ways they can stand out in a world filled with entities. Above all, a business needs to answer two fundamental questions:
- How does Google perceive my business as an entity?
- How can I optimise my own entity to improve its ranking in Google searches?
How to create a local business entity
Before you spend time thinking up ways to optimise a business entity, it’s important to answer the fundamental question of how exactly Google perceives a business as an entity.
As we saw a little earlier, the pivotal element of Google’s perception of an entity is the Knowledge Graph – a structured record of all entities that can be accessed at any time. If your business has a record in the Knowledge Graph, you can rest assured that Google has identified your entity. If not, don’t panic – but because it isn’t possible to make entries in the Knowledge Graph directly, you need to go through one of the countless sources Google taps for information. One of these is Wikipedia, though it’s almost impossible for most business owners to set up a Wikipedia entry of their own due to a lack of relevance.
For local business entities, this isn’t necessary anyway – and an entry in Google My Business does the job.
In theory, after making an entry for your local business in Google My Business, you can enter a query in the Knowledge Graph to check whether Google had identified your business as an entity. Unfortunately, Google has rather thwarted this plan for the time being because, at the time of writing (January 2020), it isn’t possible to search for local business entities in the Knowledge Graph. For example, Karlsruhe burger joint DeliBurgers has the Knowledge Graph ID “/g/11bxc3r0lx”. However, a direct search using either the name DeliBurgers or its ID returns no results. We’ll have to wait and see whether and when we’ll be able to search the Knowledge Graph for local business entities again.
An easy way to find out the entity behind each ID is to use Google Trends by entering the following URL: https://trends.google.de/trends/explore?q=[Knowledge Graph ID]
Sticking with the previous example, you’d type in https://trends.google.de/trends/explore?q=/g/11bxc3r0lx. When you do, the results show trends about the entity DeliBurgers.
How to optimise your local business entity
Creating a Google My Business entry is the digital foundation stone of your own local business entity. The next task is to optimise the entity. Generally speaking, entity optimisation can be divided into three rough areas:
- Structured data
- Visibility and consistency
- Brand building
Although a Google My Business entry is enough for Google to identify your business as an entity, it’s always important for a business to maintain a digital presence in the form of its own website. This is the only way to have maximum control of what information is displayed and how.
However, it’s also important to make it easy for Google to process the information a business publishes on its website. With this in mind, business owners should look to implement structured data. While structured data is not visible to website visitors, it does facilitate easy machine processing of the information, such as by Google. Here’s a handy introduction to structured data.
Visibility and consistency
The more relevant an entity becomes for Google, the more visible it becomes. When it comes to visibility, the Google My Business entry is a good first step – but we recommend making additional entries in other directories. The issue of which directories depends on the sector; the best options might be general platforms or more specific trade directories. If a business wants people to locate it using navigation devices powered by systems other than Google Maps, it’s worth making an entry in Navmii. Listings in the Yellow Pages or Yelp can also help to boost the visibility of a business.
The important thing is never to lose track of the entries and their purpose. All of the information entered across all online portals should be based on the same data. In the worst case, Google can punish inconsistent data by displaying a business less prominently or hiding key attributes such as its opening hours and telephone number if they’re not listed consistently throughout the Internet. Nowadays, there are products and services to help you make and manage entries in trade directories – such as List Local, which allows you to see which directories already contain listings for a given business. List Local builds on the technology developed by Uberall, which offers solutions for a wide range of sectors, such as financial services, the automotive sector and the catering industry.
Ultimately, optimising an entity comes down to the issue of brand building. Classic offline measures to set up and maintain a brand presence need to be integrated in online marketing measures more prominently than has been the case to date. Reputation management will play an important role in future, as the reputation of a business is a decisive factor in its ranking and influences its conversion rates.
Consequently, interface management is becoming an increasingly important aspect of search engine optimisation. Different teams will have to work together more intensively in future because, sooner or later, a silo mentality will have a negative impact on the visibility of a business in search engines – however good its keywords might be.
Entities have always existed. However, search engines have come to draw on the concept in order to better understand search queries – which, in turn, has changed the results they provide. Even though the topic is often discussed in theoretical terms at present, the impact of entity-based searching can already be clearly seen in some areas.
We should expect these changes to become increasingly pronounced in future and their impact much clearer. Focusing solely on keywords will just not be enough to optimise the visibility of a business in future. With this in mind, it’s important to engage with the topic of entities today to secure and maintain online visibility.
About Christian Stobitzer
Christian Stobitzer is an SEO professional at IONOS where he amongst others manages the technical SEO for the Digital Guide, an info portal about digital topics and IT hosting. Next to his position as Senior SEO Manager he has been working as freelance online marketer and tests SEO strategies in affiliate projects of different sizes.