From content management to content graph: shift to grow your business

Learn how Electronic Arts combines Contentful and a knowledge graph to reuse content efficiently, make it easier for search engines to find that content and offer recommendations that keep customers playing EA’s games longer.

Content is more important than ever

Here at Contentful, we’re constantly talking about how crucial content is to the digital world. So are our customers. Aaron Bradley’s presentation at Fast Forward, our annual customer conference, described how EA uses content and knowledge management to make content creation more efficient, target the right content to the right customers and get great business outcomes. He’s an expert on the topic — Aaron works as the knowledge graph strategist at Electronic Arts (EA), maker of some of the world’s best-loved interactive games. Here’s a recap of the most important takeaways from his talk, The evolution — and convergence — of content and knowledge management

Make information overload and outdated content a thing of the past with a knowledge base

The evolution of content management

Content management systems were originally designed to create web pages. However, we’ve moved far beyond static websites. The number and variety of devices people use to find information, shop, entertain themselves and communicate with friends and family is huge and constantly growing. Content is delivered to phones, watches, smart speakers and other devices without a visual display. For EA and companies like it, game consoles are yet another important device.

Traditional CMSes weren’t designed to populate all the applications compatible with all these devices. In order to offer the same content in multiple formats, teams using a traditional CMS manually recreate content in systems compatible with each device. In other words, they copy and paste. This method is highly inefficient, costly and error-prone — issues that compound every time content has to be updated. 

Fortunately, there’s a tool that lets you efficiently create content that can then be accessed from any device or platform your customers are using: the headless CMS.

The headless CMS and its benefits

A headless CMS doesn’t care how content will be displayed — it cares only how it’s structured. The structure of any piece of content — we’ll call it a content object — is provided by a content model that describes each element making up the content object.  

The screenshot below shows a content model that EA has created in Contentful for an article (like this blog post, for example), along with the meaning of each element. 

Screenshot of an article content structure in Contentful

A content object is made up of content elements — data. In a headless CMS, these data exist separately from how the content is displayed — that is, the presentation layer. APIs are used to access the data specified by the content model, and to provide the content in the form a particular device or application requires (the presentation).  

Using its content model in Contentful, EA can publish the exact same content to a game client such as an XBox One, PlayStation 4 or a PC game client such as the one shown below. 

Screenshot from EA's presentation of how they do content management

Because every element of a content type is defined — what the element is, and what type of data it expects — the content model has meaning as well as structure. 

In the content model shown below, every element has been defined. The text block is an element called “lead paragraph,” and it’s of the type called “rich text.” The embedded element is of an asset type called “feature art,” and it’s of a content type called “image.” This image is about the game “Battlefield V.” The entire content object is an “article” of the type “news article,” and it’s about the game “Battlefield V” and the Battlefield V map “Wake Island.” 

Screenshot from presentation describing Contentful content type breakdown

You can build even more meaning into your content by using taxonomies that denote concepts important to your business, both at the level of content type and at the content element level. Because all this information is machine-readable, it not only helps you reuse content efficiently; it also makes your content readily discoverable by search engines. You can also tap that information to perform analytics on your content. 

Content management + knowledge management = content graph

Enterprises often store their data in tables that capture details of objects or concepts that are important to the business. At EA, for example, there could be a table for each of the following:

  • Products the company sells

  • Games on which the products are based

  • The people who play the games

  • Game promotions the company sends to players

Tables can work well if you rarely need to connect different data points stored in different tables. But enterprises today generate increasing amounts of data, and if they want to be competitive, they need this data to do a lot: provide personalization, support recommendation systems and facilitate actionable analytics. Doing all this requires the ability to connect different types of data readily, and tables make that difficult. 

That’s why enterprises are shifting from table-based structures to a graph-based structure that allows data to be connected much more easily. A graph is created by describing the relationships between different objects and concepts that are important to the business, regardless of where these data reside. 

This structured representation of meaning, usually captured in a conceptual model called an ontology, makes it possible to use data more holistically. In the image below, you can see a simple illustration of how EA uses this structure.

Graph describing EA's preception of a relationship between promotions, games, products, and players

Each video game has at least one product associated with it. A promotion can be about a video game or a video-game product. These promotions are sent to players, who purchase products.

Using this conceptual model, EA can see whether an email promotion sent to a player called Mary about Battlefield 1, a game which has a product for Xbox One, influenced her to purchase that product.

Better graph of relationship between players, products, games, and promotions

Describing the relationships between different objects and concepts adds meaning to the data and provides a business vocabulary that makes it easier to use the data and get more value from it. This approach — combining a graph structure and semantically described data — is known as a knowledge graph

Combining a knowledge graph holding semantically described data with a headless CMS in which content models hold structured semantic content lets you develop a content graph. Your system now understands both the meaning of your data and the structure of your content, allowing you to use data flexibly in content products. 

Using the content graph, EA can efficiently create content in many different languages for many different markets. The game Battlefield 1, for example, is published in English, French and Portuguese. In the image below, you can see the Buy Now button in each language. That button is actually the same content element, automatically reconfigured for each language presentation. The link to the game trailer has the same structure across all languages, but the value of the target video varies by country.

Screenshot of prese tation slide with Convergence: data as content as header

To take a more complex example, EA publishes game ratings with each game. These ratings are based not on language, but on the country where the player is located. Game ratings are governed by each country’s rating agency, so the same game will have a different rating, produced by a different body, in the United States, France and Brazil — or any other country where the game is available.

Screenshot of convergence: data as content part of EA's presentation deck

It would be inefficient to recreate the appropriate game rating badge for a specific game every time a player living in a different country views it. So, drawing on the combination of structure and semantics, EA built a conceptual model of a game rating ontology to provide — both to humans and to machines — the meaning of each element of a game rating, as well as how these elements are related to each other.

Screenshot of presentation deck about data as content

From the conceptual model, EA built a logical model — a content model — to incorporate game rating elements into EA’s products: 

  • The specific rating agency that a game rating comes from

  • Information about each rating agency, stored in an external data repository and retrievable with an API as needed

  • The age range a rated product is suitable for

  • Content descriptors

  • A rating image

Screenshot of how Contentful content types are used by EA

EA uses all the data associated with a rating to generate a reusable content component: a game rating badge that can appear in the required form for any game in the company’s catalog. 

Screenshot and breakdown of the conceptual model of EA's content promotions

Data as content, content as data: increasing the value of both

By using a headless CMS with uniformly described data, EA — and any company that adopts this same approach — can use its data as content. Because the data employed in content has meanings associated with it, and because that content is well structured, the content can also be used as if it were data — in fact, that content is data.

This understanding of content as data is especially important to providing personalized content, where you need to know something about the person to whom you’re providing content, and about the content itself. 

EA’s content model for game ratings includes age ratings. The age rating itself is a content element and has its own content model, which includes: 

  • Information about each rating agency, stored in an external data repository and retrievable with an API as needed

  • The age range a rated product is suitable for

  • Content descriptors

  • A rating image

EA can combine the age-related data elements in the age-rating content model with other data elements EA has — the ages of its players — to promote the right games to the right players in a programmatic fashion. The company will never send a promotion for iBattlefield 1 to a U.S. player under the age of 17, for example, because its data on ratings shows that Battlefield 1 is rated ESRB Mature in the United States — and ESRB Mature specifies a minimum age of 17. 

There’s no need to tag the Battlefield 1 promotion with the game rating, or the age rating, or the minimum age for the game. All that data is already in the content model for the promotion, and it’s all connected via models. This makes it easy to send out the promotion to the right people.

Building a recommendation system 

EA has other information about players, such as what they’ve purchased, what they’ve been playing, and what device they’ve been playing it from. By using this player data in content models, EA built a recommendation system that resulted in 40 percent more game sessions — a great business result.

In the image below, you can see a snapshot of the news section in the lobby of a game, displayed in a gaming system. It’s an uncurated collection — maybe the items here will help a player succeed and enjoy the game more, and maybe they won’t. 

Screenshot of how recommendations look in EA's client

EA knows quite a bit about players of Battlefield 1: the game modes they favor, the character classes they’ve played, and where they’ve hit an impasse in the game. The company also knows what content it has that’s relevant to Battlefield 1.

Taking the API-first approach that characterizes a headless CMS, EA launched a recommendation system that drew on its data about a player’s progress and its knowledge of EA videos offering tips to players that help them play the game better. Rather than an undifferentiated river of news, EA displayed videos to each player offering the tips they needed to break their impasse and keep playing. 

Screenshot of EA's client

In a test of the system, EA recorded a 40 percent increase in session days for players who received content recommendations over those who did not. Because the individualized recommendations improved their playing experience, players enjoyed the game more and played longer.

Wrapping it up

The recommendation system EA built is just one example of how using a headless CMS can help any company get better business results. In a nutshell, a headless CMS offers these benefits:

  • Content can be efficiently created, reused and reconfigured for different purposes 

  • The ability to use models to build personalized content for recommendation systems

  • Content created with a headless CMS is easily discovered by search engines

  • You have a foundation for rich content analytics

But Contentful is even more than a headless CMS. It’s an entire content platform that lets you seamlessly integrate other products into your content hub. Read more about how the Contentful content platform empowers digital teams to build digital experiences for the future. 

About the author

Don't miss the latest

Get updates in your inbox
A monthly newsletter to help you build better digital experiences with Contentful.
add-circle arrow-right remove style-two-pin-marker subtract-circle