The future of campaigns has arrived. Predictive audiences with machine learning in Google analytics.

In recent years, you will have heard new technologies applied to websites and applications to help customer service and to create newsletter campaigns. Of the first, a clear example is the bots that have been able to serve you on many of the websites of today. These, through programmatic rules, manage to solve doubts to users and thus save time, save costs and improve conversions. On the other hand, email automation campaigns allow you to send clear and direct messages to your audiences depending on how they have interacted on your website. As, for example, send an email to a user who has left products in the cart, but has not finished the purchase.

All these advances allow you to improve the conversion rates and with it the income you get from your ecommerce or corporate website. However, technology goes further and is allowing us to anticipate user behavior through predictive analytics with the data you collect on your system. A clear example of this advance is the technology called “Machine learning”: what this technology does is a study of current data and learning through the new data it receives. After a training period it allows you to start executing new actions.

A clear example is the new audiences that Google has created in its web analytics tool. It has been able to create a system that learns from the behaviors that users carry out on your website. In addition, it creates audiences of people that Google anticipates that they will have the same behavior as the people analyzed above. There are 2 types of audiences:

  • Predictive Purchase Audience: This audience collects a list of users that you expect to buy within the next 7 days.
  • Predictive Audience of Abandonment: Gets the probability that users will not page in the next 7 days.

audiencia

With these audiences we can attack with specific messages to finish closing a sale or to try to recover the user. But in order to get access to them you must have e-commerce configured:

  • The tool must collect at least data from purchase events
  • These data must have a minimum audience so that the tool can learn from their behavior and make more accurate predictions. 1,000 users are required to activate the relevant predictive rules (purchase) and 1,000 who do not.

This technology allows you to remarketing the users that enter your website, but with a greater probability of success since these users are based on very specific intentions within the conversion funnel. Although for the campaign to be more effective, you must prepare your copywriters and a good strategy for offers or a list of products of interest.

WE WANT TO HELP YOU
TRANSFORM YOUR BUSINESS

We turn your brand into the
reference of your industry

WE SPEAK?

Leave a Reply