Understand What Is The Difference Insights
- Navigate to a story’s What is the Difference insights and explore them.
- Understand the relationships between variables.
- Use these insights to help maximize customer lifetime value (CLV).
In Einstein Discovery, What Is The Difference insights are comparative insights that help you better understand the relationships between explanatory variables and the goal (target outcome variable) in your story. These insights, based on a statistical analysis of your dataset, help you figure out which factors contribute to the biggest changes in the outcome variable. Einstein Discovery uses waterfall charts to help you visualize comparisons in What Is The Difference insights.
By isolating an explanatory variable, you can see and learn how it relates to the whole, and how it compares to another explanatory variable. For example, you can compare sales performance for manufacturing customers to your overall sales performance. In addition, you can compare sales performance between manufacturing and distribution customers. Finally, you can add a filter to focus on a smaller slice of your data (such as a particular sales region).
In this unit, we use What Is The Difference insights to explore the story you created previously (see "Create a Story" in the "Einstein Discovery Classics" module). Recall that the goal of this story is to maximize customer lifetime value (CLV). CLV is a metric that predicts the profitability over the entire lifetime of the company’s relationship with a customer.
Select the What Is The Difference Insight Type
On the Insight Navigation bar, click the down arrow in the upper right, and then click What Is The Difference.
The Insights Navigation bar displays this category but no graph. To see a graph, you must first select a variable.
Compare an Explanatory Variable with the Global Average
In our example story, the global average represents the CLV of all data in the dataset. It is useful to compare the CLV of a single variable with the global CLV average. To select a variable, for Relating to (select a variable) on the left of the Insights navigation bar, select Industry - Shipping.
When the calculations are complete, you see the most statistically significant insights in a waterfall chart.
CLV appears at the top of the graph as a reminder that we configured this story to maximize CLV as our outcome variable.
At the top of the chart, the gray bar labeled Industry is Shipping (Average) shows us the average CLV when Industry is Shipping. To see more details, hover over the gray bar.
When our story maximizes CLV as the outcome variable, a red bar in the chart shows a condition that reduces CLV from the average. Hover over a red bar in the graph.
In this example, when Industry is Shipping and BillingState is Texas, CLV is -89, or 89 below average. Notice that the corresponding insight description on the left is highlighted in gray.
When our story maximizes CLV as the outcome variable, a green bar in the chart shows a condition that increases CLV from the average. Hover over a green bar in the graph.
In this example, when Division is Standard Hardware, CLV is 190 above average. Notice that the corresponding insight description on the left is highlighted in gray.
The blue bar at the bottom of the chart shows the Global Average (Outcome), which represents the average CLV for all data in the dataset (20,136).
Compare Two Variables
Next, we add a second explanatory variable and compare the two. On the Insights navigation bar, go to Between (select a variable) on the right and select Industry - Technology.
When the calculations are complete, you see a waterfall chart comparing the two industries.
At a glance, this chart shows that CLV for Industry is Technology (Average) outperforms Industry is Shipping (Average) in many ways. For example, when accounts are rated hot, Technology has a better lifetime CLV than Shipping. But in Texas, Shipping has a much better CLV than Technology.
Hover over the gray bar at the top of the chart that shows Industry is Shipping (Average).
Hover over the blue bar at the bottom of the chart that shows Industry is Technology (Average).
Comparing the actual CLV numbers confirms that the average CLV for Technology is higher than for Shipping.
Add a Filter
Optionally, you can add a filter to further focus your analysis on a subset of the data. On the far right side of the Insights navigation bar, click Search story insights and choose Type - Consulting.
When the calculations are complete, you see a waterfall chart comparing the two industries with only Consulting data.
In this example, Small Terms and Division is Standard Hardware show the highest correlation for maximizing CLV when Type is Consulting.
In this module, you continued your work as the VP of operations for a major automotive supplier. You dug deeper into the story you created in the Einstein Discovery Data Integration module. You learned how to interpret several of the insights that Einstein Discovery uncovered in your data. You looked at descriptive insights that detailed what happened, and at comparative insights that show what is the difference when comparing variables. The story was filled with insights about your data. Exploring the story helped you discover relationships between the CLV of your accounts and other variables that can influence CLV.