Experience Personalisation: Data and Content
Experience Personalisation Series, Part 2 : Content
This article is the second in a three-part series, you can read part 1 here: Experience Personalisation: Getting the Data Right
“Data + Content = Personalisation.” This is the simple formula we use here at Adobe to express the importance of knowing who you’re serving content to, and when and where to serve it. The better you’re able to personalise that content—and its delivery—the more effectively each piece of content will help drive conversions.
In this series of three articles, I’m exploring all three aspects of this equation.
In the first article of the series, we examined two crucial ways of gathering data on customers: building progressive profiles from multiple interactions with the same users; and pulling in actionable profile data from other systems throughout your organisation, such as your customer relationship management (CRM) system, and internal systems like enterprise resource planning (ERP) and data warehouse (DWH).
Now, in this second article, we’ll be taking a deeper dive on content.
Automation
In the most basic terms, automation is the use of algorithms and machine learning to model customer behaviour. One of the most powerful uses of automation is automated behavioural targeting—leveraging large amounts of customer data and machine learning to determine which variables in a customer profile will be most predictive of a purchase.
Any marketing system you use should provide a comprehensive solution for modelling customer behaviour—a solution that takes both the actions and preferences of the general population into account, as well as those of the individual visitor. This way, changes in user behaviour on your site caused by external factors—for example, a new product is released, a star is pictured wearing your clothing or using your product, or the season changes—can be picked up quickly; and when an individual exhibits strong preferences that override the responses of the general population, this can also be captured.
For example, say you run a site for a pay-per-view TV operator. You need to make customers aware of the television channels you provide, as well as your sport and movie offerings. You see a particular visitor multiple times—and this visitor always browses the cricket pages in the sports section, as well as looking at their account online. From this information, you can construct a picture of who this customer is, based on the individual’s account profile as well as on the types of content that they like to consume.
Then, when the new football season starts, and users of your site start responding in great numbers to the football content you’re serving up on the homepage, your system will know that this particular user is more likely to respond to cricket-related content—and rather than wasting a serve, your system will present that user with a homepage tailored around his interest in cricket. Meanwhile, visitors for whom we don’t have this level of detail will see a football-related creative, because that’s what the general population of the site are engaging with.
Sky TV in the UK recently used this exact approach to deliver relevant content and messaging to users engaging with the Sky Shop, in order to drive incremental sales and upgrades on desktop and mobile.
Real customer profiles, of course, include many more elements than this—and the more robust your customer profiles, the more useful and time-saving variable targeting will be.
Optimisation
Adobe Target’s automated behavioural targeting capabilities integrate seamlessly with the other content development and delivery tools throughout Adobe Marketing Cloud—meaning that as soon as the system discovers a highly predictive variable, it can immediately begin assembling the ideal creative to target that variable, and serve that ideal creative to the customer across channels and devices.
Another aspect of automation comes through the use of something we call auto-allocation. In a standard A/B test, you have to run the test, collect enough data in terms of responses or conversions, read the results and then implement the winning variation—which will often be different per segment or audience that you have on your digital properties.
With auto-allocate functionality, the solution automatically determines which experience is working best for your given measures of success—then serves that experience automatically to your visitors, so that you maximise the exposure of that best performing message, banner, or creative.
Even better, other experiences are also being continually tested—so if any significant variable changes, the solution can react and change the experience for you, without the need for manual intervention.
Recommendations
We’ve all seen the recommendations at the bottoms of pages on sites like Amazon—similar and related items that customers viewed and bought. Those recommendations provide a tiny window into a vast collection of crowd behaviour, which the solution is using to make recommendations to you, the individual user – who may not even have a profile in that database yet.
These types of recommendations are a popular example of the use of behaviour data from the general population to predict how a user may behave. In other words, the more people in the world are engaging in a certain activity at a certain time—going caravanning in the summer, for example, and reading about it on your site—the more likely your customer is to imitate the trend. However, a great recommendations solution will also enable you to bring individual preferences and behaviour into account—just as we saw with automated behavioural targeting.
When you bring these two types of automated analysis—automation and recommendations—together, you’ll gain insights not only on what your customers want and need, but also on what they plan to do soon, as well as what they’re likely to do in the near future. This level of insight enables you to serve personalised, relevant creative like never before.
In the third and final article of this series, I’ll be delving into the personalisation spectrum, and exploring how data and content come together to drive increasingly personalised creative. See you there!