Strategy Choices for Content Personalization with Sitecore Experience Platform
Personalization has been a buzzword in the marketing technology landscape for years. While a wide range of ready-to-use tools for personalization already exists, there is still too little thinking about the overall process for delivering consumer-tailored content on websites, mobile apps, emails and globally any digital touch point.
Why do I need personalization? What do I expect from personalization as a brand? What would my consumers expect from a personalized experience? These are only a few of the questions marketers must ask – and answer – before they start looking at personalization as a performance optimization tool. Those questions make even more sense for large-scale digital properties with lots of traffic and lots of content.
Sitecore Experience Profile offers the possibility to address personalization from the marketing strategy standpoint. As one key component of the Sitecore Experience Platform, this powerful personalization engine enables marketers to reflect part of their digital strategy in Sitecore via a visual interface while encapsulating most of the technology complexity behind the scene. In many cases, setting up a strategy-driven personalization could be achieved mostly through configuration in the CMS with little code development.
With this in mind, implementing personalization with Sitecore should always start with some key strategy considerations. One of the main concerns should be “What is my primary marketing objective when implementing personalization?” The solution may differ depending on the answer to that key question. The first approach is to look at it from a consumer expectations standpoint, while the second approach is more focused on marketing goals.
Approach 1: Promoting Content Based on its Compatibility with Consumer Expectations
Dividing content into different categories and arranging those nicely in a navigation menu is nothing new. But it still comes with some limitations – for one, consumers often have to browse through different sections to discover content they may like.
Moreover, the way content is visually organized on a website, for example, is more driven by the website global content strategy than by different consumer interests. While content strategy and consumer interests can sometimes match, websites typically implement at least one secondary categorization method. This can come to life as tags or topics that appear at the bottom of a piece of content and lead to a listing page that normally cannot be reached via the regular navigation.
With that in mind, let’s now look at a good basis to define consumer interest:
- Content category
- Cross-category tags assigned to the content
- Any potential explicit consumer data captured
How Does This Come to Life with Sitecore?
Now let’s consider a digital property for which the primary objective of personalization is to serve each consumer with the content they may be most interested in. Sitecore offers an easy way to define multiple-interest profiles that can be computed against each consumer as they browse through content. Then, it becomes simple to promote content based on its compatibility with the consumer’s expectations.
With this approach, users are exposed to content that makes sense for them without having to find it. As this does not require additional coding on top of Sitecore Experience Platform, it is a relatively quick win for marketers. However, this process relies strongly on the quality of the content tagging. A couple of things could go wrong:
- The application of interest profiles to content could be done poorly.
- The content selected for tagging may not be the most relevant.
- The interest profile definition could be challenging if the content strategy is not clear enough.
There is only one key principle to follow: Keep it simple. Usually, defining very few profiles with very few possible values (three to four) is the best course. The easier the profile is to define, the easiest it will be to decide which tag to apply to content, and the more reliable the consumer behavior profiles will be.
Approach 2: Promoting Content Based on Its Proven Ability to Drive Marketing Goals
While the first approach provides a better experience for users, it does not necessarily bring any optimization from the business perspective. What if marketers want to leverage personalization to optimize efficiency or ROI? Sitecore also makes this simple with the use of marketing goals and content value per visit.
A brand’s website, for example, usually has a handful of key objectives for its visitors. A cart checkout, a feedback post, a brochure download, a lead capture – these are a few common marketing goals for a website. Generally, a website would combine a few of these common goals with other goals. Then, the user experience would be designed to optimize the number of such transactions. While it is true that the user experience is a key driver of success, content also holds a critical role in influencing consumers. Dedicating significant time and resources for content strategy and content performance analysis is already a no-brainer for marketers. Looking at how content performs over time and continuously readjusting content strategy is already a standard practice. What if it were fairly easy to address at least a part of this process with automated personalization via Sitecore?
Content Performance Basic Principles
For this second approach, let’s consider a website with a few marketing objectives such as:
- Capturing a lead via a registration form
- Receiving a request for a brochure
- Receiving a service or product request
In this context, personalization could also be used to promote content based on its proven ability to drive the above goals. To get there, we must identify content that has previously played a role when a visitor performed such an action. If we can identify such content – and can track each occurrence of the desired actions – we can easily figure out which content is playing a significant role in the marketing strategy. In other words, it would be feasible to confirm how helpful each piece of content is for the marketing strategy. These content performance principles are old news and have been applied to analytics data for years.
This whole process could be even more accurate if we now consider that a product request is a more engaging action compared to a brochure request. If we effectively define how much we care for each of those actions from a marketing standpoint, the content performance concept could be brought to a higher level.
Sitecore and Content Performance Automation
The way content performance is usually leveraged is iterative: Content strategy is set up, analyzed, updated, analyzed, updated … and so on.
The same “rinse and repeat” principle could be applied via Sitecore in a more automated and systematic way. Sitecore enables marketers to reflect their marketing goals in the system. Then, by technically linking each goal to different user actions (this could be as simple as clicking on a link), the goals would automatically be triggered as consumers perform actions. The interesting part starts with the content scoring process. To put it simply, each piece of content that contributed to a goal-related user action would be rewarded with engagement value points. Those engagement value points will help confirm which content is more involved in goal-related actions.
Capitalizing on its easy way to compare content performance, Sitecore offers the possibility to define personalization rules that would look at content performance as one way to decide which content to display. Typically, it would be easy to specify the following rule: “On the homepage, always promote the best-performing content on the site in a banner.” The great benefit with Sitecore is that this whole process happens in real time. Thus, content performance is continuously re-assessed as visitors consume content and reach goals. As a result, performance-based personalization rules would always be relevant based on the immediate reality. That best-performing content on the homepage banner could vary from day to day with no human intervention, based solely on visitor behavior.
There is one fundamental difference between this option and the profile-based option: With this option, content selection is fully driven by consumers’ activity. There is no need to take assumptions up front about which content should be displayed to which audience. Of course, that takes away some control from the brand. But if the content being displayed has the highest chances to convert the visitor, why would you want to show something else?
Bear in mind that this approach comes with a few watchouts:
- The content used for personalization should also be promoted outside personalization areas, with no personalization mechanics. All content should have more or less equal chances to trigger goals; otherwise, the whole content scoring principle is distorted.
- The content performance-driven approach seems to be more relevant with scale, while the profile-based approach could easily be applied even with only a few pieces of content to recommend.
Naturally, the next question is: Why should we choose between profiling and content performance? In short, there is no such limitation. Mixing both principles, it is not too complex to promote content based on its proven ability to drive marketing goals for previous consumers that have the same expectations. It could be achieved either with a bit of coding or with machine learning.
While we are yet to discover what Sitecore 9’s Cortex has to offer, there are numerous options for coupling Sitecore personalization functionality with machine learning. Microsoft Cognitive Services is one. But if we start dealing with machine learning, the possibilities are even bigger. It is typically fairly simple to promote content that performed the best for consumers who came through the same traffic source.
Jonathan BOBO, Chief Technology Officer for Proximity IO and Paris