When it comes to customer experience optimization, the focus is on the exact approach. But here speed often competes with accuracy. Therefore, companies have to weigh carefully and choose between a very fast but possibly less data based personalization (e.g. purely session based) or a slower but more data based alternative (e.g. considering the entire customer journey and purchase history).

An extract from our new whitepaper, in which four prominent top experts from the fields of online marketing and digital analytics answer questions on the topic of “Customer Experience Optimization”, offers an essential decision-making aid.

Speed or Accuracy

To answer this question, André Morys from konverionsKRAFT sets priorities:
“The faster a good customer experience is delivered, the more ROI is achieved. So why wait? From a conversion optimizer perspective, we want to influence a customer’s behavior. The earlier the customer perceives this stimulus, the earlier the behavior can be changed”.
But it is also important to consider the other perspective for a qualitative understanding of the customer experience. André Morys continues: “On the other hand, of course, we also have requirements for a certain quality. Showing the customer nonsense because too little or bad data is available is not good either. So you have to find the optimum under these conditions. I always say that good results need high-contrast changes. And that’s why I plead for ‘the sooner the better’. So real time.”

traffic sign stop city

Marc Preusche from LEROI (part of DEPT), can second this by highlighting how important real-time personalization is, especially for new customers / new visitors who don’t bring any data history or other audience data: “If a completely new user downloads your app directly and there is no information where he came from, it will be pretty difficult to execute any personalization without real-time processing of the currently given data”.

Michael Böhme from Zalando sees differences in the speed of different business processes, which are also fundamental for a good CX, and certain personalization features:“Important business processes have to be analyzed and optimized in real-time. One example is the checkout and payment process, in which we use real-time data and machine learning models to prevent fraud. On the other hand, there are onsite features that can use one day old data to personalize certain experiences.”

In his experience, when considering whether real-time data or a rule-based, rather heuristic approach makes sense, one should decide individually for each application: “A good heuristic approach sometimes works well enough. It really depends on the application. Not every function / experience must and can be optimized by machine learning and real-time user data. It is about what is your value proposition (in terms of CX) and where to invest to gain a competitive advantage. It is important to understand how the current solution and possibly a better technical solution will increase customer satisfaction and how much impact different approaches would have”.

“Hybrid” solutions between rule-based and fully automated data-driven personalization are a common intermediate step, especially for larger, growing companies. Marc Preusche sees it this way: “Rule-based seems to be more controllable, at least in the beginning. Before you switch to complete data-driven personalization, you are always dependent on a set of rules.

But later you come to a point where you think: ‘What can I do with the real-time component anyway? As soon as you realize that the real-time component has only marginally more added value than the non-real-time component, you can look at the hybrid with regards to an evolution towards a completely automated, data-driven personalization.”

……

Would you like to know what other insights our experts share about CX optimization and data-driven personalization? Then don’t hesitate and download our free whitepaper NOW!

The 3 highlight topics of DMEXCO 2019
4 Myths of Customer Experience & why you shouldn't believe them