This article is the fourth in a series about AI and how you can begin to think about and leverage the various AI technologies.  Last time we talked about the different functional categories of AI applications as a foundation for more detailed discussions about the AI features offered by the Salesforce platform.  Today we will take a look at the older of the AI capabilities - Predictive AI.  Predictive AI has been a part of the Salesforce platform for the better part of a decade, beginning with the announcement of Einstein back in 2016.  Remember, predictive AI refers to the ability to make predictions or forecasts about future events or outcomes based on historical data models and machine learning algorithms.

For a lot of the features that we look at below, Salesforce has already created models to support the data analysis and results.  However you also have the ability to not only adjust those models in some cases, but to also build your own models using the Einstein Prediction Builder.  This gives you, the user, the best flexibility to tune the Einstein engine for your specific needs.  Remember that large amounts of data are needed to properly train models.  If you don't have enough data, in some cases predictions may not be shown but in other cases, Einstein will instead use a global model which is built and trained with anonymous data from many Salesforce customers.  The documentation for each feature goes into more detail on how the models actually work.

The general process followed for all of these predictive models is first the data is prepared - it is cleaned, transformed and organized to make it suitable for ML algorithms.  Einstein can use this data to identify patterns and relationships in the data set.  The most relevant variables or attributes, known as features, are then selected from the available data.  The models are then trained on historical data.  Once the model is trained, it can generate predictions for new data.  Note that Einstein models are designed to adapt and improve over time.  As new data becomes available, the models can be retrained to ensure their predictions remain accurate and relevant.

With that introduction out of the way, let's dive in and take a look at some of the more popular features. Einstein in 2023 is a foundational element of the Salesforce platform and its features can be found throughout the different products on the platform.  Some of these features are included automatically, while others are add-on features with additional licensing cost.  Your Salesforce Account Executive can provide more details on which are included, as well as the cost of any add-on features.  Note that this is not an exhaustive list of features utilizing the Einstein engine, but some of the more used features. 

 

Einstein for Sales

  • Einstein Forecasting analyzes your team's past opportunities to create a model that uses details about the opportunities and related accounts, history and related activities, along with owner historical trends, to predict how much your sales teams will sell at the end of a forecasting period.
  • Einstein Lead Scoring determines conversion patterns by analyzing your past converted leads, and uses this information to identify which of your current leads have the most in common with those prior converted leads.  This is converted to one or more scoring models which helps identify which leads are most likely to convert.
  • Einstein Opportunity Scoring analyzes your team's past closed opportunities to build a scoring model which helps predict which opportunities are most likely to be won.

 

Einstein for Marketing

Marketing Cloud Features

  • Einstein Engagement Scoring predicts consumer engagement with email and mobile messaging, using customer data and ML algorithms to assign scores for the likelihood of each contact to engage with emails or interact with push notifications.
  • Einstein Messaging Insights alerts you to changes in your marketing performance by keeping you informed as to how your email sends and journeys are performing.  Einstein monitors the engagement rates, such as opens, clicks and unsubscribes, of your email sends, and when an outlier result is found, an insight is generated with information about the anomaly and its context.
  • Einstein Send Time Optimization uses ML to predict optimal send times so that a user is likely to engage with your message.  Einstein uses each contact's latest engagement data for commercial sends to recreate its sending model on a weekly basis.

Account Engagement Features

  • Einstein Behavior Scoring uses ML to uncover the most influential behavior signals across past and current prospect engagement, helping to identify prospects whose behavior suggests that they are ready to buy.
  • Einstein Campaign Insights identifies data related to engagement activity, content, and audience characteristics to provide real-time insights about your running campaigns.

 

Einstein for Commerce

  • Einstein Product Recommendations helps personalize the shopping experience by automating product discovery on your storefront.  The feature suggests personalized products based on customers’ past purchasing history and current behavior on your site, and also uses shopping trends to recommend popular items.
  • Einstein Commerce Insights captures shopper, order, and product data and displays products that are most commonly purchased together, to identify your customers' habits and purchasing behavior.
  • Einstein Predictive Sort calculates the affinity of an individual customer to your products and automatically tailors search and category pages based on every action a shopper makes.  The personalized sorting experience is continuously refined for each shopper as more and more data is collected.

 

Einstein Next Best Action

Einstein Next Best Action allows you to use rules-based and predictive models to provide anyone in your business with intelligent, contextual recommendations and offers, and those insights are then surfaced directly within Salesforce.

 

Einstein Discovery

Einstein Discovery helps augment your business intelligence by using statistical modeling and machine learning to identify, surface, and visualize insights into your business data.  Einstein Discovery is integrated into your Salesforce and Tableau environments so that you can quickly integrate enhanced data analysis and predictions into your operational workflow without any custom coding.

 

Einstein Prediction Builder

Einstein Prediction Builder is a simple wizard that helps you quickly build custom prediction models without any coding.  You can create predictions for any part of your business - sales, service, marketing, commerce, etc.  Once your models have been created, trained, and deployed, your developers can utilize the Einstein Prediction Service REST API to embed your predictions into any web site or application.

 

As you can see, there are a lot of different ways to start taking advantage of Predictive AI on the Salesforce platform.  You can learn more about the Einstein features by visiting the following link on Trailhead:  https://trailhead.salesforce.com/content/learn/trails/get_smart_einstein .  Trifecta's experts are also available to talk more about these features and help you understand how to best take advantage of them.

Scott Geosits

CTO, Trifecta Technologies