How should Sitecore customers prepare for machine learning

The Sitecore Symposium in Vegas last November was an eye opener on a lot of different levels.  The preview of Sitecore 8's functionality & polish during the keynote got everyone in the audience excited for the future of the product. Nestled in the product keynote were previews of some game changers that are expected to evolve over the next few years.  Some of those include; 

  • The aggregation of customer sessions across devices/channels, and the ability to view that in the form of the Sitecore xFile and timeline
  • The incorporation of Sitecore originated personalized content, multivariate experiments, etc. to externally managed content, like ad network copy, PPC messaging, etc.
  • Machine learning, and its impact on customer experience

The focus of this message is the notion of machine learning in the future of customer experience management.   

Machine learning surfaces patterns that we often times can't see ourselves.  Each discrete customer interaction is tracked and evaluated through machine learning.  The result is a continuum of recognizable patterns that humans are not suited to uncover.  We're not suited because we don't have the attention span, ability to discard our biases or the necessary time to surface these patterns.  The inventory of recognizable patterns are then used to predict customer behaviors, at scale.  What might we learn from machine learning? 

  • Based on real-time, in-the-moment customer actions, how likely they are to convert from call to action A verses call to action B
  • Whether a customer has a higher probability of attrition (imagine beginning a win-back campaign before the customer actual leaves)
  • How to pair the customer with the sales rep or customer service rep that's most likely to connect with him/her. 

In our data-driven future, marketers will likely spend more time tuning their organization's customer experience model; their algorithm.  What does this mean for the CMO or CDO?  It means they are going to have to add 'data scientist' to their ever expanding quiver of required skills.  Algorithm driven business automation is not new.  You've heard the eHarmony ads; 29 dimensions of compatibility... Uber has an evolving algorithm that drives their 'surge pricing' variable.  UPS has an immense algorithm that controls its driver mapping & delivery sequencing.  If you look around, you can find this level of self-learning automation everywhere. 

So, let's bring the conversation back to what machine learning means to me, as a Sitecore customer.  First, this is still an emerging topic and not fully mainstream yet.  Take a look at minute ~ 1:08:00 of Michael Seifert's keynote at Symposium last year to get a good feel for how Sitecore views machine learning.

As we move towards machine learning becoming a key part of customer experience management we'll see that a solid Sitecore architecture is a requirement.  Well-designed customer personas and customer journey stage definition are necessary to provide the 'who' and 'where' inputs.  A clearly documented taxonomy, tied to profile & pattern cards help make the necessary associations between content, visitors, browse paths, etc.  Identified major outcomes, goals along the way, and relative importance (engagement value) provide insights into what success looks like. 

The same inputs that help marketers today will help machines tomorrow.  

If you'd like to keep current with Sitecore related posts, please consider following me on Twitter.