Discover Einstein Recommendation Builder

Learning Objectives

After completing this unit, you'll be able to:

• Explain what Einstein Recommendation Builder is and what problems it solves.
• Describe the benefits of Einstein Recommendation Builder.
• Identify use cases that can benefit from Einstein Recommendation Builder.

Recommendations Enter the AI Age

Recommendations. They’re everywhere.  

That clerk at the shoe store suggesting running shoes. Your movie streaming service proposing the latest detective series. Your doctor prescribing that you eat more vegetables and less pizza. 

But not all recommendations are equal. Consider this scenario. You’re taking a date out to dinner, and you ask the waiter for advice on selecting a cheese board. He suggests one with an impressive selection of cheeses, but it comes at a hefty price of $100, leaving less money for dessert that could melt your date’s heart. His second recommendation is the basic cheese board for $10. Unfortunately, these cheeses are dull, and you suspect that your date will question your judgment if you pick the alternative. 

Each recommendation has its positives, but neither is a great recommendation.  

What if that waiter had access to a treasure trove of information, such as comparative demographic data, on which to base his recommendation? Or, what if he knew which cheese your date usually orders, which cheeses are popular, or which ones have been sent back by diners?

What if he had access to Einstein Recommendation Builder?  

It’s Einstein’s World. We Just Live in It.

Einstein Recommendation Builder combines the power of AI and the utility of big data to not just make recommendations, but to make the best recommendations.  

Einstein Recommendation Builder works by accessing the data stored in your Salesforce CRM. Einstein identifies similar targets, such as comparable accounts or analogous customers, by analyzing thousands of data points and attributes from your data, such as revenue and industry, to come up with intelligent recommendations.  

You can recommend anything (Recommended Items) to anyone (Recipients) by connecting two Salesforce objects. You can connect any two standard or custom Salesforce objects (including managed package objects).  

Here’s an example. Suppose you have a work order from a customer, but it doesn’t have the right product parts to resolve the customer’s issue. With Einstein Recommendation Builder, you can find out which parts were used to resolve similar work orders in the past.

1. Choose a Salesforce object to recommend (the Recommended Items object). In this case, it’s a Product object, which contains parts information.
2. Then, choose a Recipient object. That’s the object that receives the recommendation. In this case, it’s a Work Order object.
3. Finally, choose an Interactions object, which stores past interactions between these two objects. In this example, it’s a Product Consumed object.
4. Name your recommendation and click Save. Einstein Recommendation Builder is off and running, using the history of similar products and work orders to recommend the right parts for this job.

Here are other examples of how you can use Einstein Recommendation Builder. (Objects can differ.)

A recommendation still has to become an action, but that’s easy. Using recommendation strategies, you pass recommendations on to Einstein Next Best Action to turn them into actions, such as offering a 10% discount on the new version of the customer’s mobile phone. With Next Best Action, you can add business rules to your AI recommendations to meet your business needs—for example, to ensure there’s sufficient inventory before offering this product promotion to your users. Accepting a recommendation triggers a flow, such as one that sends an email to a customer, offering the discount, or starting an application for a new credit card. User actions/responses are recorded in Salesforce, and Einstein uses this to adjust its recommendations for the future. That’s right, Einstein Recommendation Builder gets smarter.

You can display recommendations on product detail pages, console apps, or within Salesforce communities, as well as in external applications via a REST API.

Top Reasons to Use Einstein Recommendation Builder

There are a number of reasons to use Einstein Recommendation Builder.

You get the best recommendations. Einstein Recommendation Builder is powerful. It can process far more data than people can. This power, combined with sophisticated AI, provides high-quality recommendations.    

It’s automation, baby. Sure, you could go look at a client’s purchase history, revenue stream, company size, industry, and the kind of potato chips they have in their vending machines, and then formulate a plan. Einstein Recommendation Builder does that heavy lifting for you. 

Clicks, not code. There’s no coding necessary to create a recommendation with Einstein Recommendation Builder, and very little for the whole end-to-end process. (Some coding may be required. To create a flow, for example.)  

Transparent metrics and do-overs. Einstein Recommendation Builder shows you the quality of the recommendation and explains the numbers behind the number. Don’t like the quality of a recommendation? Adjust some parameters with segmenting and field exclusion and run Einstein Recommendation Builder again to get a new one.  

Scalable. Currently, if you only have 5 or 10 products, you can train your sales reps to recommend the best one for each customer. But what if you have a large product catalog? And as you add more sales reps, it gets harder to train everyone to have the same expertise on all your products. Einstein Recommendation Builder can help you scale, by consistently picking the most relevant products from thousands of choices. And it works equally well for the newbie sales rep who just joined the team last week.   

Okay, Sounds Great, But How Do I Use It?

We go into more detail about using Einstein Recommendation Builder in the next unit, but here’s a brief overview.  

Like ancient Gaul, using Einstein Recommendation Builder is divided into three parts. (Unlike ancient Gaul, it involves no pitched battles between Roman legions and Celtic tribes.) The three steps are:

1. Build. Start Einstein Recommendation Builder, choose from one of the pre-configured templates or pick your own objects for custom recommendation, name your recommendation, and turn the crank.  
2. Evaluate. After Einstein Recommendation Builder has created the recommendation for you, you can either deploy it or fine-tune your criteria and run Einstein Recommendation Builder again to improve it.
3. Deploy. After you have a recommendation that you like, you pass it on to Einstein Next Best Action, and display it wherever you like (for example, in a Console app). 

And that’s Einstein Recommendation Builder. Simple and powerful. 

Resources

• Salesforce Help: Einstein Recommendation Builder
• Video: About Einstein Recommendation Builder