The internet has come a long way since Tim Berners-Lee first laid the foundation for a working web back in 1990. The amount of content users now create and interact with has grown exponentially, and the profound effect it has had on the ways we consume information is impossible to ignore.
Attention spans are dwindling and people have become very discerning and impatient online. We have basically trained our brains to scan and immediately reject content that is not relevant to our needs. To combat this issue and keep readers’ attention, we use content personalization to make users’ website experiences more tailored to their interests.
What is Content Personalization?
Content personalization can take many forms, but it is generally intended to deliver better value and more relevant content to users, help them find the content they need, and make them convert quickly. A personalization strategy can be based on rules or machine-learning algorithms, or both.
Rules-Based Content Personalization
As a basic personalization method, rules-based content personalization uses a number of simple, manually created and easily adjusted rules that, with a number of personally identifiable attributes, divide your audience into smaller segments. The segments can then be individually targeted and sold to.
Think of rules-based personalization as a series of IF-THEN statements. Enriched with AND / OR operators, they allow us to create a more fitting experience for each user group based on the location, language, and other data collected during users’ previous interactions with the website.
Predictive Content Personalization
Predictive content personalization, also referred to as machine-learning personalization, is the more advanced and AI-driven way to dynamically display the most relevant content to each user.
Unlike the rules-based method, it does not target whole segments; instead, users are identified at a more granular level, and a more personalized website experience is created for them. It puts more focus on displaying content and messages to users based on their intent, rather than just on the readily available information about their interests and previous behavior.
Ask us anything about predictive content personalization
What Data Does Predictive Content Personalization Use?
Machine-learning personalization uses a combination of algorithms, filters, and analytics. It either “knows” or “predicts” users’ typical behavior on the website, their favorite product categories, sorting methods, and more. To do this, machine-learning personalization utilizes:
- Basic algorithms that dynamically recommend different items without using any personally identifiable information about the users. This can include showing them new products, current promotions in the store, trending posts or products, or products currently browsed by other people.
- Advanced algorithms that further customize the content to each user by the available personally identifiable data or demonstrated behavior.
For example, based on demonstrated behavior, the algorithms will assign each user to a group of users with similar preferences (think providers of streaming media like Spotify or Netflix). The algorithm dynamically predicts other products or content they might like, saving them the hassle of rummaging through everything that’s not exactly their cup of tea.
Algorithms can be used to create decision trees that are most likely to lead to a conversion, individually for each client.
Filters allow companies to further tweak the results of algorithms, and make them exclude or include particular elements.
How Data-Management Platforms Power Predictive Content Personalization
Predictive content personalization heavily relies on vast amounts of data about each user’s interactions with the website. To this end, the data has to be aggregated from various sources within a data-management platform (DMP). The content-personalization platform can then be employed to leverage this data to help a website boost conversion rates and make the personalization more accurate.
The DMP, once populated with sufficient amounts of data from desktop and mobile devices, can crunch the information to deliver a more detailed image of each user. This involves, among other things, merging and segmenting the data sets using various factors.
Before that happens, however, the DMP has to sync first-party cookies (from the same website) with a number of second-party cookies (from other websites), creating continuously updated segments of data. In this way, the DMP lays the foundation for successful content personalization.
This process allows companies to refine the segments of users visiting their website. Then, depending on whether additional technologies, like device fingerprinting and cross-device tracking, are in place, the result is an unparalleled segmentation, impossible to achieve with traditional methods.
We Have the Data. What Now?
With the DMP data at hand, a content-personalization engine is needed to properly leverage the information and display the user’s relevant content—recommendations, promotions, and messaging that make the experience more personal.
Examples of Predictive Content Personalization
There are plenty of examples of predictive content personalization around the web. Since the algorithms usually work under the hood, the average user may be completely oblivious to the fact that they even exist.
The most prominent examples include:
The large catalog of movies and shows makes it virtually impossible to create a single home page experience that would be equally fitting for each of its 57+ million subscribers. The service prides itself on its elaborate content-customization algorithms, which populate the home page with the most relevant content for each subscriber.
The generation of home-page content starts by finding groupings the member is most likely to enjoy, all based on the information the website knows about them. The recommendations are dynamically updated, as the taste of the viewer (and the way they navigate through the website) may change with time. On top of that, there are dynamic page metrics or maturity-rating filters blocking specific kinds of content.
Music-streaming service Spotify’s algorithms let people discover new music by using a number of personalized playlists, like Release Radar, Discover Weekly, and Spotify Radio, using previous listening habits to predict new music users will like.
The website’s most recent personalization effort is its Time Capsule playlist, which assembles a list of 30 throwback songs “to take you back in time to your teenage years. This will, naturally, be different for each user.
Given Amazon’s vast product database, it would be hard, if not impossible, to make the front page in any way relevant to each shopper, and since predictive personalization goes hand in hand with ecommerce, Amazon’s focus on content personalization is not unfounded—as much as 35% of its yearly revenue comes from the recommendation engine.
The most obvious personalization strategy involves displaying customized recommendations of products to clients on the front page of the store. It can be used to display discounts and exclusive promotions to loyal customers who have previously made a purchase in the store.
Challenges of Predictive Content Personalization
On the flip side, content personalization may not be for everyone. There are a lot of challenges on the way to success. Implementing a recommendation system that would be valuable for your audience will certainly involve a lot of experience with A/B and multivariate testing methods. Netflix explores these challenges extensively in one of their blog posts on Medium.
Before even considering a foray into predictive content personalization, it is good to start small with the rules-based method. Why? There is an array of challenges inherent to personalization like insufficient data to tap into (typical of small companies), inability to properly activate the data at hand, and incompatibility of various sources of data.
The imminent implementation of the General Data Protection Regulation in May 2018 is bound to cause some stir in the industry as well. The directive completely redefines cookies and the way they will be used by MarTech solutions. This will certainly impact the effectiveness of personalization.
Predictive content personalization allows marketers to harness the power of machine learning and improve the experience for each user visiting their sites. Sure, it dramatically increases the chances of conversion, but there is more than meets the eye. Real-time recommendations also lend a helping hand to the users, streamlining their experience and reducing time to find what they need.