How to deliver a truly personalized user experience by making intelligent use of your available data – in real time and automated

There was a time when I struggled to identify broader use cases for real-time analytics. Although real-time analytics was touted everywhere, my experience was different: It was rather the exception that there was such a need for instant data comparison or analysis and decision-making in real time. An example of such an exception were “data war rooms” during the Super Bowl or similar peak events. Here you had to manage massive, but very short-term advertising campaigns. I can recall many discussions within the analytics community, most of which had no real use case even for very sophisticated and advanced teams. Data granularity and data accuracy was much more important. And of course, how to set up a “source of truth” and effective data visualisation and interpretation.

That was years before I was with ODOSCOPE. Meanwhile, the technology and especially e-commerce has evolved a lot, and analytics has been automated impressively (which still doesn’t mean it works on its own!). Data driven real-time decision making is not more rocket science. Analytics is not more only looking at the “data from yesterday”. That is still the case, of course, but the analytics data is also used right away for providing a better customer experience.

What does personalization in e-commerce mean?

A simplified view on personalization is this: A user who showed interest for X or tends to buy products from Y, or belongs to the customer segment A is treated that way for his/her next visits or touchpoints and  likely in parallel for behavioral targeting. This approach is somewhat similar to social media bubbles. They are constantly preaching to the converted, or feeding and re-feeding one bubble endlessly. We all know this is stupid, sometimes even dangerous, but after all effective. Further, it can be expensive when targeting the already converted or simply uninterested users.

Situations matter

Marketing lives a lot in persona-thinking. This is a long time approved model which helps a lot to direct the right message to a target group. User profiling is more or less derived from there. However, when it comes to a single and concrete situation on which a decision must be made, it misses an important point: “How are you today?” A question like this is basically the starting point for any human conversation. The answer to this question helps a lot to understand the very customer situation and mood, regardless of the customer background in terms of purchase history, customer segment, demographics, etc. Much more important is the situation. The same user is driven by his/her very situation, depending on the time of the day, the device type, the region etc.. The same person is acting and reacting very differently in different situations.

A great customer experience must not reflect all historical behaviors and all user details in order to serve well. It needs to fit to the moment and the actual desire. Why should a sales person in a store know a bunch of personal information (yes, it is!) about you just to help you finding the right jeans?

Personalization in e-commerce

Personalization in e-commerce is much about which product is displayed first or how an overview is sorted, what recommendations are made and best-guessing (basically betting on) what is relevant to the user at that very moment. Personalization of the user experience on your own website or in your own online shop does not require spending advertising money. It is also less fraudulent by default.

It is possible to automate personalization very accurately. Yet, for this you do not need to build user profiles by collecting any possible personal information just to better guess if red or blue is the favorite color.

Apparently, you have all the data and knowledge to do that on your own. You have the data of the situation. 

Data-driven automation for effective decision making

A bunch of anonymous but very useful information comes with every single web visit by default. Even without having a tracking consent there’s always the http-protocol from which several anonymous data points are provided per default. This includes e.g. device type, day and time, (accurate enough) geo-location, and many more. There’s a lot you can do with this already – and in real-time. In addition to this, a user is constantly sending information when clicking on a category, using filters, surfing between overview and product pages.

What’s needed is taking the current visit data points (use on-the-fly, no saving or storing) and map those against the historical web-analytics and product-order data. Here, you have granted consent anyway. ODOSCOPE’s lightning fast real-time cluster immediately identifies significant correlations and peer groups. Groups of similar users in that very moment are like statistical siblings of the current web visit. The identified statistical sibling’s conversions are pointing to the most relevant content which is now displayed to the specific user. All this happens in some milliseconds during a page load. Every visit is treated individually, in real-time and served with automated decisions. These decisions are significant and relevant to the user in this very moment.

This is data-driven automation at it’s best. It is safe, smart and super performant.