Use Stories to Get the Big Picture
- Create a Tableau CRM dataset and import data into it.
- Use your imported data to create an Einstein Discovery story.
- Describe the order in which insights are presented in a story.
- Describe the types of insights that a story gives you.
In this module, we learn about creating stories, navigating the insights that Einstein Discovery generates from stories, and how to interpret those insights. Imagine that you’re the VP of operations for a major automotive supplier, where you’ve developed a reputation for ace detective work. The executive team was so impressed when you solved the case of the shrinking margins in the Einstein Discovery Basics module. They’re back, asking you for more help. Now they want more information about the customer lifetime value of accounts.
What is Customer Lifetime Value?
Customer lifetime value (CLV) is a metric that predicts the profitability over the entire lifetime of the company’s relationship with a customer. Looking at CLV can help you find the group of customers who are potentially the most profitable. That way, more marketing resources can be allocated to them.
What is a Story?
You use Einstein Discovery stories to analyze data in Tableau CRM datasets. Einstein Discovery uses a story to perform a comprehensive statistical analysis of your data powered by statistics, machine learning, and AI.
- helps you uncover relationships between a business-relevant outcome (sometimes referred to as the outcome variable) and the explanatory variables that are potential influencers of that outcome
- has a goal of minimizing or maximizing the outcome (for example, maximizing margin or minimizing expenses)
- specifies the dataset columns to include in the analysis, along with other preferences
- is filled with insights about your data, accompanied by visualizations and narratives to help you better understand the insights that Einstein has uncovered
- numeric values that represent quantifiable measures, such as revenue, discount, cost, or duration
- text fields with two values (binary outcomes), such as win or loss, churn or retain, or approved or rejected
In our case, CLV represents a numeric outcome. Let's create a story to work with.
Try Einstein Discovery with a Developer Edition Org
Before you work through this Trailhead module, sign up for a free Tableau CRM-enabled Developer Edition org. This org is a safe environment where you can practice the skills you’re learning.
Download the Data
Before we can create the story used in this module, we need the data to analyze. Download the CSV file called AcquiredAccount.csv and save it to your computer.
The CSV file has 11 columns. It contains one row of information for each of the 10,000 different companies that our auto parts manufacturing company does business with. Here is what the first few rows of the CSV file look like:
Create and Populate a Tableau CRM Dataset
The next step is to get the data from the CSV file into a Tableau CRM dataset.
- From the App Launcher ( ), find and select Analytics Studio.
- Click Create | Dataset.
- Choose CSV File as the source for your new data.
- In the file-selection window that opens, find, then select the AcquiredAccount.csv file you downloaded, and then click Next.
- Accept the defaults and click Next.
- Accept the defaults and click Upload File. Tableau CRM creates a dataset and imports the data from the CSV file.
Create the Story
Now, you’re ready to create a story from this dataset. Begin by telling Einstein Discovery which outcome variable to focus on. In this module, we want our story to maximize the CLV variable.
- While in the dataset settings screen, click Create Story. Tableau CRM Analytics Studio launches the Story Setup wizard.
- In the first screen, for I Want to Analyze, select CLV as the outcome that you want Einstein to analyze.
- Accept all other defaults and click Story Type.
- In the Story Type screen, click Insights & Predictions, then click Setup Options.
- In the Setup Options screen, select Manual and click Data Options. Notice that the selected outcome variable is first in the list. The remaining columns represent explanatory variables. An explanatory variable is a variable that you explore to determine whether, and to what degree, it can influence the outcome variable for your story. In the Correlation column, Einstein shows you the percentage by which each field is statistically correlated to the outcome. The Division field has the highest correlation. However, Account Id has the second highest correlation. Let’s stop and think for a moment. Does an arbitrarily assigned account ID have any influence on CLV? Probably not. We know that because we know our business. In this case, statistical significance does not translate to real-world significance. Therefore, let's remove this field from our analysis to speed up the analysis and get clearer results.
- Clear the check box next to Account Id.
- Click Create Story.
When it’s done analyzing your data and discovering insights, Einstein Discovery shows you a results summary.
Here are the key areas of the interface for Einstein Discovery stories:
|Story Headline||Name of this story, selected goal, most recent version.|
||Shows you the list of variables in your story and their correlation to the story outcome.
||Tools you can use to view predictions and improvements, update story settings, and other tasks.
|Story Version Summary||Summary of story insights, including version comparison.|
|Insight Summary Panels||List of variables, ordered by correlation, that positively or negatively impact a story.
Insights in Our Story
The story presents the insights that Einstein Discovery uncovered for you. The first insights you see are descriptive insights, which tell you more about what happened according to the historical data in the dataset. Use descriptive insights to explore, at an overview level, the factors which contribute to the story's outcome.
Order of Insights in the Story
Insights are shown in order of statistical significance. The first insights that you see are the ones that explain, statistically, the most variation in the outcome variable. The insights that appear later, as you scroll through the story, explain variables that, statistically, account for less of the variation in the outcome variable. At the beginning of an analysis, we don't presume cause and effect. The insights show us statistically significant patterns that give us clues about where to dig in for more detail. You are free to explore the insights in any order you want.
Other Types of Insights
- comparative insights
- diagnostic insights
- predictions and improvements
This module covers descriptive and comparative insights. The Einstein Discovery Story Insights module covers diagnostic insights, as well as predictions and improvements.
Dive Into Our Story Insights
Now you have an overview of stories and insights. Let's dive in to see how to get the most out of each type of insight in your story. The next unit explores descriptive insights.
- Salesforce Help: Explain, Predict, and Take Action with Einstein Discovery
- Salesforce Help: Create Datasets from Uploaded CSV Files
- Sample Data to Download: AcquiredAccount.csv
Rights of ALBERT EINSTEIN are used with permission of The Hebrew University of Jerusalem. Represented exclusively by Greenlight.