Get to Know Einstein Discovery

Learning Objectives

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

Describe Einstein Discovery capabilities.
Explain the types of use cases Einstein Discovery addresses.

What Is Einstein Discovery?

Einstein Discovery enables you to augment your business intelligence with statistical modeling and supervised machine learning in a no-code-required, rapid-iteration environment. Use Einstein Discovery models to quickly surface insights in your business data and to predict future outcomes. Deploy an Einstein Discovery model to inject machine learning based recommendations across your organization. For example, use Einstein Discovery predictions in your workflows or add Einstein Discovery predictions to your Salesforce pages to have Einstein suggest ways to improve the predicted outcome. For an overview of all that Einstein Discovery can do for your organization, check out Einstein Discovery: Quick Look.

Note: Einstein Discovery requires either the CRM Analytics Plus license or Einstein Predictions license, both of which are available for an extra cost.

What Kinds of Use Cases Can Einstein Discovery Help With?

Einstein Discovery supports these common use cases for business outcomes.

Use Case	Applies To
Regression	Regression for numeric outcomes represented as quantitative data (measures), such as currency, counts, or any other quantity. For example, Einstein Discovery can help you with the monetary value of your opportunities.
Binary classification	Binary outcomes for text data with only two possible results. These are typically yes/no questions that are expressed in business terms. For example, Einstein Discovery can help you win opportunities.
Multiclass classification	Outcomes with 3 to 10 possible results, represented as text data. For example, a multiclass model can help you predict the most likely next stage for an opportunity: whether it proceeds to the next stage, goes back to a previous stage, or even skips a stage.

Illustration showing the three supported use cases using opportunity as an example.

What Business Outcome Do You Want to Improve?

Successful journeys begin with the first step. For Einstein Discovery solutions, the first step is to select a business problem you want to solve. In your business, survey key performance indicators (KPIs) that could benefit the most from deploying an Einstein Discovery-powered solution. The business outcome must fit with one of the supported use cases: regression, binary classification, or multiclass classification.

In this module, we explore a sample scenario in which the goal is to maximize opportunity wins. The business outcome is either win or lose. Therefore, we’re using Einstein Discovery to solve a binary classification problem.

How Do You Implement an Einstein Discovery Solution?

Illustration showing the sequence of tasks to complete in this module.

This module takes you through the steps commonly followed to implement an Einstein Discovery solution. Each unit addresses a different step in the process, from building a CRM Analytics dataset to exploring data insights and predicting and improving outcomes.

As you become familiar with the tasks involved with each step, you begin to understand the iterative nature of implementing a successful Einstein Discovery solution. Einstein Discovery is designed for rapid exploration, experimentation, and iterative improvement.

Progress is cumulative, not linear. You learn as you go. Every step of the way, you use built-in feedback to check your results, review your assumptions, ask new questions, make adjustments, and try again. Clean up your data. Add or remove columns in your dataset. Apply filters and transformations. Tweak the model threshold. And so on. As you fine-tune your approach, each improvement can lead you closer to better operational outcomes.