Start tracking your progress
Trailhead Home
Trailhead Home

Understand Predictions and Improvements Insights

Learning Objectives

After completing this unit, you’ll be able to:
  • Navigate to a story’s Predictions & Improvements insights and explore them.
  • Identify the best probable future outcome for your scenario.
  • List the factors behind a prediction.

Understand Predictions and Improvements Insights



The instructions in this unit assume that you have successfully created an Einstein Discovery story according to the steps in "Create a Story," the first unit in this Trailhead module.

Predictions and Improvements insights help you explore what might happen in the future. For example, you can interactively perform “what if” analyses in your story. Einstein Discovery provides you with predictions and suggested improvements based on a statistical analysis of your dataset and predictive analytics. To help you visualize these insights, Einstein Discovery uses:
  • waterfall charts for predictions
  • bar charts for suggested improvements

These predictive insights help you make smart, calculated decisions about what can happen in future transactions. Einstein Discovery also suggests actions you can take to improve outcomes. While other insights show existing data from past transactions, Predictions & Improvements insights calculate statistically probable outcomes of future events. Predictions are similar to the kind of regression or machine-learning analysis that data scientists conduct using advanced analytics tools.

A prediction is just that: an anticipated outcome based on Einstein Discovery’s statistical analysis of your data. Of course, predictive insights aren’t guaranteed business outcomes. However, they can help you investigate and understand factors that possibly influence the business outcomes that interest you.

Understanding Prediction & Improvements

The Prediction & Improvements insights show you the most important details behind the prediction for a single goal, which in our case is CLV.

Get a Prediction for a Single Variable (Division)

Let's start by seeing how Einstein Discovery can help predict CLV by industry.

  1. On the Story navigation bar, click Predictions & Improvements.

    Predictions and Improvements insight type on the Story navigation bar.



    You must select at least one variable to see a predictive insight.

  2. In By Changing (select a variable), select Division.

    Select the Relating to variable.

    Einstein Discovery generates a list of insights. The first insight shows the Predicted CLV by Division.

    Predictions by Division.

This graph shows us which divisions yield the highest predicted CLV (Raw Materials and Mapping), and the lowest (Standard Hardware).

Investigate Raw Materials.

Raw Materials appears to be a lucrative division. Let's investigate further. In Relating to (select a variable), select Division - Raw Materials.

Division is Raw Materials.

Let’s examine this graph.

  • Baseline represents the average CLV of all fields combined.
  • Division is Raw Materials represents the predicted CLV for Division is Raw Materials. The bar shows that the predicted average CLV is $483 above the baseline.
  • Expected Impact of Other Fields represents the predicted average CLV for other fields (an additional $954 above Division is Raw Materials). That’s a very significant impact!
  • Predicted Outcome represents the predicted CLV for the selected variable (Division is Raw Materials) plus the expected impact of other fields. Hover over the blue bar to see details. The predicted outcome for this scenario is $23,903.

Add Another Variable (Industry is Manufacturing).

Let's predict how Raw Materials works in combination with a second variable. In Relating to (select a variable), select Division - Raw Materials and Industry - Manufacturing.

Filtered by Division is Raw Materials and Industry is Manufacturing.

Einstein Discovery refreshes the insights based on this filter.

Division is Raw Materials and Industry is Manufacturing.

The graph shows that Industry is Manufacturing by itself lowers CLV by 77 and the combination of Industry is Manufacturing and Division is Raw Materials lowers it by 415. That’s significant! However, when you add back the other categories (Division is Raw Materials and the Expected Impact of Other Fields), the net effect is that the Predicted Outcome is 22,505. Manufacturing lowers CLV in this scenario.

Explore a Different Variable (Industry is Apparel).

Next, let's try a different industry - Apparel - to see how it might affect the predicted CLV outcome. In Relating to (select a variable), select Division - Raw Materials and Industry - Apparel.

Division is Raw Materials and Industry is Shipping.

Einstein Discovery refreshes the insights based on this filter.

Division is Raw Materials and Industry is Apparel.

The combination of Division is Raw Materials and Industry is Apparel boosts the predicted CLV to 25,240.

By trying different variables and combinations of variables in what-if scenarios, you can use the power of predictive insights in Einstein Discovery to help you uncover ways in which to improve business outcomes.


In this module, you learned how Why it Happened insights give you a deeper understanding of the complex relationships in your existing data. You saw how related and unrelated factors affected the observed outcome, then used Predictions and Improvements insights to make better decisions about future business actions.