WalkMe AI Predictive Analysis

Updated on January 19, 2018
Download PDF

Use artificial intelligence (AI) to reduce your customer churn, increase adoption, and drive conversions. WalkMe’s AI Predictive Analysis (AIPA) feature analyzes user behavior data to predict future actions. This group of users can then be targeted for customized campaigns, helping you reach your business goals.

The Short Version

AIPA surpasses traditional analytics by providing predictive insights into user intentions and proactively driving users toward the business goals most relevant to them.  AIPA works on both traditional websites, via the Rule Engine, and on mobile applications, via campaigns.

Here are just a few of the many use-cases through which AIPA can take your business to the next level:

  • Use predictive insights to prevent user churn through timely interactions addressing user needs.
  • On mobile apps, increase the engagement with your campaigns by identifying users’ “Happy Moments” when they are mostly like to complete your desired goal.

How It Works

The AIPA solution is made up of two distinct processes:

  • Data collection: WalkMe constantly collects and analyzes behavioral data from all the users in a given site.
  • Likelihood analysis: Statistical analysis (derived from the data collected) that predicts the likelihood of churn or other actions to take place in the user’s next visit to (or absence from) the website/application.

AIPA’s capabilities currently differ between WalkMe for Mobile Web and WalkMe for Mobile Native Applications.

Website AI

Once WalkMe gathers sufficient data (for high-volume websites/apps this can take just a few days), AIPA begins progressively refining its data and using the results to identify a group of users who are likely to churn.


WalkMe defines “churn”  as the point at which a customer will not return to a website or application. The data points collected to determine likelihood of churn are centered around two components: First, on a given website, and the way and frequency in which its end users visit and utilize the site, and second, on the end users themselves and their likelihood of churn based on their interaction with the given website. These data points are aggregating constantly, and, like any statistical model, AIPA’s model gets more accurate as more data points are gathered.

User Behavior Sample Size

Once AIPA is turned on, the time it takes before the algorithm’s accuracy reaches the desired 85% threshold of accuracy can range from a few days to a few weeks, depending on user traffic and variance.

The value add comes from the way AIPA predicts who will churn and allows you to engage them in any way you see fit to prevent this: Upon page load, AIPA compares a particular user to the characteristics of the group, and, if the user is flagged as likely to churn, AIPA will initiate any actions (ShoutOut Auto Play Rules are currently supported) that you’ve specified should occur once the rule is evaluated as “true.”

Mobile AI

AIPA for mobile apps works similarly to the web version, except that it can identify two additional user groups:

  • Those likely to perform a positive action relative to a pre-set campaign.
  • Those likely to complete a pre-set business goal.


Contact your CSM to activate AIPA.


Related Articles