Chapter 6: Declarative Customization Options

This will be a hands-on chapter that shows how you can use generic Einstein declarative features to create your own solutions as well as discussing when that is the right approach. By working through this chapter, you will gain knowledge of how and when to use the declarative features of the Einstein Platform. We will be using our Pickled Plastics Ltd. scenario throughout to ground our examples.

In this chapter, we're going to cover the following main topics:

• Introducing Einstein declarative features
• Giving timely advice with Einstein Next Best Action
• Predicting outcomes with Einstein Prediction Builder
• Generating insights with Einstein Discovery Stories

After completing this chapter, you will have gained an understanding of how to use out-of-the-box features to configure advanced AI solutions on Salesforce, using clicks not code.

Technical requirements

To follow along with the examples in this chapter, please register an analytics-enabled developer org. This can be requested using the form here: https://developer.salesforce.com/promotions/orgs/analytics-de.

The Code in Action (CiA) video for the chapter can be found at https://bit.ly/3iztq7g.

Introducing Einstein declarative features

In this chapter, we change gears again. So far, we have primarily been looking at features that come out of the box with various Salesforce clouds. We've had to do some configuration, but nothing particularly strenuous. Mostly, things work out of the box.

While that can be a great strength, it also means that when you hit the limits, they are hard. Typically, you can't bend a feature to match requirements, even though, technically, the requirements are quite close to the core functionality.

Once you hit the limit, your next port of call should be to see if you can meet your requirements using some combination of declarative features to avoid the overhead of having to use code and train your own models. Powerful as such solutions might be, they are also more expensive, higher-maintenance, and riskier than declarative solutions.

In this chapter, we will go through three features of the Einstein platform that can help you create...

Giving timely advice with Einstein Next Best Action

In this section, we will cover Einstein Next Best Action, which, technically speaking, is a framework for making in-context recommendations for record-related actions in Salesforce. That means based on a strategy and the properties of a record, you can give customized recommendations.

In the next two sections, we will first give an overview of the feature and then move on to a simple hands-on example.

Overview of Einstein Next Best Action

Einstein Next Best Action is best thought of as a framework for providing in-context action recommendations on Salesforce record pages using a highly configurable recommendation engine. Typical use cases include the ability for customer service agents to give special offers, upsell, or cross-sell options to customer's when they have them on the phone, but it can also be used, for instance, to ensure data quality by prompting field inspection or to suggest actions that may lead to the...

Predicting outcomes with Einstein Prediction Builder

Einstein Prediction Builder is often referred to in Salesforce documentation as custom AI for admins. This statement has some claim to truth, at least if you have well-populated data objects with fields that have relatively stable correlations between them. In the following sections, we will review the capabilities of this feature and attempt to build a basic prediction ourselves.

Overview of Einstein Prediction Builder

Einstein Prediction Builder works by analyzing historical data in Salesforce objects using machine learning under the hood. While we don't have access to the details of the model used, we could perhaps assume that it uses the same kind of model tournament to determine the optimal algorithm that we have seen in previous features.

While there is no data science degree required to configure Einstein Prediction Builder, you do need to define your use case and success metrics upfront, if you want to use...

Generating insights with Einstein Discovery stories

Einstein Discovery is a part of the Tableau CRM suite that Salesforce provides for advanced analytics. Tableau CRM is a topic that deserves several book-length treatments in its own right, so we will only be able to scratch the surface here. However, because of the brilliant machine learning capabilities inherent in Einstein Discovery stories both from an analytical and predictive standpoint, we will cover as much as we can in the next section and then follow with a small configuration example.

Overview of Einstein Discovery

Einstein Discovery is an add-on to Tableau CRM that adds statistical modeling and machine learning to the analytical capabilities of the core product. It shares the ability with Einstein Prediction Builder to generate automated machine learning models on the basis of datasets, but because it is embedded in Tableau CRM, it can make use of the data wrangling tools found in that product, which overcomes many...

Summary

In this chapter, we have looked at declarative customization options for Salesforce AI. We have learned about Einstein Next Best Action and seen how it enables us to give in-context recommendations that can improve the way our users work with Salesforce either using simple rules or by generating advanced recommendations based on AI.

Then we looked at Einstein Prediction Builder, the proverbial "custom AI for admins" that can generate good predictions for fields on your objects, at least if you meet the general conditions. This feature can generate predictions easily but only works well if you have a lot of data with relatively stable relationships between the fields.

Finally, we had a very brief look at Einstein Discovery, a feature that deserves a much longer treatment than we can give it here. This is possibly the most powerful of the declarative features in our arsenal, but it does require Tableau CRM, which will disqualify it from being used in many organizations...

Questions

1. What are the four key elements of Einstein Next Best Action?
2. Where can the data used for prediction by Einstein Prediction Builder be placed?
3. How can predictions from Einstein Discovery be activated?