The Blog

  • March 7, 2013
  • KXEN — Predictive Modeling for Marketing, Sales and Service

    This post is part of an occasional series on the AppExchange as Salesforce.com celebrates the seventh anniversary of its launch.  The series will focus on some of the most interesting AppExchange applications of the last year.

    KXEN is a ten year-old company focusing on cloud-based predictive modeling that helps Salesforce users to build templates based on statistical models that predict business activities in sales, marketing and service, such as lead quality in sales.  For enterprise companies with thousands or millions of potential customers, predictive data modeling is a low cost and fast way to identify the next best opportunity, activity or lead.

    Predictive modeling seeks to build a probability-based representation of reality that it then compares with specific new instances to arrive at a forecast of an outcome.  Some of the great online retailers rely on relatively simple predictive modeling when they suggest additional purchases based on prior customer purchases as well as those of other similar people.  That’s what’s behind offers that start with, “People who have bought this have also expressed interest in that.”

    The difference between predictive modeling and BI or analytics, is that analytics is essentially backward facing, it reports on data collected and stored in data warehouses.  Predictive modeling analyzes situations and scores them to produce probabilities that can be applied in various new situations.  Analytics tells us what happened and modeling tells us what might happen.  Most businesses need both approaches for different reasons and predictive modeling has a lot of attention driven by big data..

    Predictive modeling is one of the few application areas that still does well as an on-premise solution in part because it is computationally intensive but also because many predictive modeling solutions still require “power users” to get the models right.  But KXEN saw an opportunity in providing cloud-based predictive modeling solutions and decided to make it available through the AppExchange.  One of the company’s strong points is that it has made predictive modeling accessible by a non-technical user.

    KXEN’s Predictive Lead Scoring uses both native and custom fields and each model is specifically built for a salesforce.com customer.  The model instantly scores sales leads by analyzing leads that have converted in the past into qualified sales opportunities.  Over time, the model continues to add to its experience data, which keeps it updated with changing market conditions.

    Customers simply install the application in their Salesforce instance and KXEN starts delivering optimal lead scores with the same sophisticated predictive engine that KXEN also provides to some of the world’s largest banks, telecommunications providers, retailers and e-businesses.

    KXEN offers predictive apps for all three Salesforce Clouds.  Predictive modeling has many uses in the Sales Cloud beyond lead scoring including outbound call scoring and customer segmentation.  In the Service Cloud, its predictive models can indicate next best activity, predict churn and function as a recommendation engine for e-commerce.  In the Marketing Cloud, predictive modeling is a boon to campaign targeting, email opt out and a recommendation engine for social content.  And it is very useful in win back campaigns.

    Predictive modeling is an essential component of CRM in industries that scale to very large customer bases such as telecommunications, retail and financial services but it is also important in smaller organizations where speed and accurate decision-making are important, for example in online retail.

    KXEN was founded in 2002 and it has more than five hundred customers worldwide.  Given the nature of this solution and the demand for analysis driven by big data it is no surprise that KXEN was one of the most successful companies on the AppExchange in 2012.  Given those market dynamics this trend should continue, and you don’t even need a predictive model to see that.

    Published: 5 years ago


    Comments are closed.