Salesforce says it’s added new machine intelligence features to its marketing cloud that will let marketers predict customer behavior.

The first feature, which Salesforce calls “Marketing Cloud Predictive Scoring,” allows companies to score customers according to the likelihood of their taking steps toward some positive behavior — like downloading an app or making a purchase.

Another feature, called “Predictive Audiences,” allows companies to generate a list of customers with a certain score, and then devise a special plan for marketing to them. For instance, the company might isolate a group of customers who are exhibiting signs of falling out of touch, and undertake a marketing plan to win them back.

Salesforce said that companies Life Time Fitness and Rue La La are now using the predictive intelligence features. But Salesforce also told VentureBeat that it’s a bit too early to show proof of success.

A screen shot of the Predictive Audiences feature.

Above: A screen shot of the Predictive Audiences feature.

The Salesforce platform uses CRM data, marketing data, and relevant contextual data (such as web browsing activity and email engagement) to create the predictions. The data is analyzed in real time, and the predictive engine can adjust its engagement recommendations at each customer interaction.

Meghann York, director of product marketing at Salesforce Marketing Cloud, compares the predictive tech to GPS or Waze (the social driving app) — technologies that gather real-time data to help drivers understand the various twists and turns on the way to their destination and suggest actions, such as taking alternate routes.

Salesforce’s predictive technology is aimed at companies that use digital channels to market on a mass scale, York said.

Salesforce says Predictive Scoring and Audiences will be included with the Salesforce Marketing Enterprise edition, and sold as an add-on to Pro and Corporate editions. Pricing starts at $125 per month.