Predictive personalization: is it good or bad? In one of our previous articles we’ve already remembered the movie "Minority report", even though it was quite another topic, the one dedicated to Creative sandbox and more specifically to augmented reality.
In this movie people in the Subway pass along large wall-sized screens, special system renders each individual giving them something special or exciting.
In 2002, when the film was shot, this type of advertising sounded quite Sy-Fy, the product of a distant future, but today, 11 years later, predictive personalized advertising is absolutely real.
Predictive personalization is defined as the ability to predict customer behavior, needs or wants - and tailor offers and communications very precisely. Social data is one source of providing this predictive analysis, particularly social data that is structured. Predictive personalization is a much more recent means of personalization and can be used well to augment current personalization offerings...
...discover highly relevant content requiring minimal effort to find. Both bet that people still value content and try to serve up good stuff for those moments in which they have nothing else to do. Both attempt to provide a machine-augmented curated media experience.
People use search engines, communicate in social media websites, create custom content...Everyone of us creates a profound trace on the web, and it makes pretty easy to predict our interests or needs.
That’s exactly what Amazon did for the last couple years trying to make use of Big Data. Let’s take Google Now as an example, this service is pretty familiar to owners of Android smartphones. The App analyses your account activity and reports most essential info on the screen, including traffic jams statistics, exciting places nearby, game score of your favorite team, exchange rates, weather in your area, and much more.
JWT advertising agency made a survey for Americans and Englishmen. People were asked to comment on their attitude towards the analysis of their personal data, made by major companies, analysis that allows creating personalized marketing offers.
It was quite logical: the elder respondent was, the more intense is the fear of being exposed. The most curious thing was that 64% of users are willing to abandon their privacy for the sake of their own purses.
Those of marketers who see the future in personalized ads should not fear. While predictive advertising helps consumers save money, such ads will be called for and will be really effective.
Now let’s have a look at some practical ways of applying predictive personalization
Provide relevant content
Website personalization simplifies the search of the content. This technology adequately identifies key characteristics of each visitor, and categorizes them based on predefined rules. At the same time, visitors feel that website is personalized and enjoy the benefits of "noise reduction" (irrelevant information) and see only most interesting content.
Create targeted ads
This technique means building up ads depending on customer’s needs, based on the history of their interaction with the website. It sufficiently increases customer satisfaction and leads to an increase of conversions. While the history of user interaction with the site accumulates, this data can be used to develop unique, relevant offers in future, as well as group visitors with similar interests.
All personalization techniques can be divided into two categories:
- Rule-based and user segment personalization.
- Personalization based on predictive analytical algorithms.
Rule-based and user segment personalization. This type is based on the rules i.e. the practice of using history data, behavioral data and environmental data for creating unique proposals based on those predefined rules. Typical personalization rule takes the following form: "If a visitor makes a follow-up, show the X offer."
One example of easily tracked and segmented client’s characteristics is geographic location. If a customer visits the site selling cloths in New York, user will be offered personalization ads based on his IP address and will see coats and jackets, but if the IP address belongs to Las Vegas they will be offered sandals and slippers.
Personalization based on predictive analytical algorithms. This presumes the use of mathematical systems to monitor visitor behavior to develop predictive models and deliver most relevant content for each visitor. In contrast to the targeting strategy based on rules, algorithmic targeting creates and connects larger, and potentially infinite number of computer-generated micro-segments all of which develop when the model learns.
An example of this technique is a bank website which serves a number of different customers and offers a wide range of financial products. In this case targeting is based on algorithms that are effective in connection with information based on behavior of customers. All of these things are used to predict most effective individual ads for each visitor + a large number of potential options.
Some statements to conclude
Be sure to install web analytics system to track results. In order for personalization to work properly and have a positive effect on conversion and business, you need to measure effectiveness of targeting efforts for designated purposes.
In addition, it is desirable to use tools for creating multiple targeting levels. You need to determine the possibility of using targeting within your websites and use relevant techniques for every situation. In circumstances where these technologies can act independently without causing conflicts, marketers may use different targeting techniques together.
And the most important thing: you need to manage actively all aspects of targeting. In behavioral targeting there is no such option as "set and forget". All targeting efforts should always be checked at regular intervals and periodically compared to the control group (which was not personalized to verify the effectiveness of your efforts).