My Einstein Prediction Builder Toolkit
Do you want to create your first Salesforce Einstein prediction? Having trouble getting started? You’re in luck! We’ve created a Toolkit that will equip you with the skills needed to create your first prediction using Einstein Prediction Builder. This kit comes equipped with four tools, which are available in downloadable format here. To use the Toolkit, visit the link and create a copy of the Slides for yourself. This Toolkit will guide you through the steps of creating your prediction, from the first step of figuring out what you’ll predict through the last step of creating a holistic, big-picture view of your predicted scores.
As an example of how to use the Toolkit, we’ll put ourselves in the shoes of Cloudy’s Computing Co., a SaaS company that sells licenses on a per user per month basis. We’ll go through the journey of how Cloudy uses the Toolkit to create their first prediction.
Tool 1: Prediction Storybook
The first step in creating your Einstein prediction is to determine what you’ll predict and to come up with strategies for how you’ll use the prediction to drive benefits to your company. Our first tool, the Prediction Storybook, will help with that. Similar to a Mad Libs book you may remember from your youth, this Storybook gives you complete control over how your prediction story plays out. All you need to do is fill in the blanks!
These are the key concepts the Storybook covers:
- Set Persona: Who will use the prediction? What is their role at your company?
- Identify Pain Point: What is a problem your business is facing? What pain-point is it causing? All predictions should overcome a pain-point, so make sure you identify a pain-point it can overcome before continuing
- Set Prediction: This is what Einstein is going to predict. Make sure you can relate it to Salesforce!
- Benefit Realized: This is the benefit the prediction will bring to your company. This will overcome the pain-point identified earlier
- Action Statement: All predictions should drive action to be taken. Consider what your co-workers would do if they were armed with the knowledge of the prediction.
Let’s take a look at Cloudy’s Storybook; recall that they are a SaaS company that sells licenses on a per user per month basis.
Revisiting our key concepts, let’s look at how Cloudy used the Storybook to set up their prediction.
- Set Persona: Sales Reps will use prediction
- Identify Pain Point: Customers are churning, leading to lost revenue
- Set Prediction: The prediction will measure churn likelihood
- Action Statement: By knowing a customer’s likelihood to churn, Cloudy will be able to take action by offering discounts to Customers at high risk of Churn.
- Benefit Realized: Through customer-saving actions like offering discounts, Cloudy’s customer churn will decrease, ultimately increasing revenue
By completing the Storybook, Cloudy has learned they will use Prediction Builder to predict Churn Likelihood. Now, Cloudy is ready to use the next tool, the Building Blocks, which will help them create the prediction in Salesforce.
Tool 2: Building Blocks
After finishing the Prediction Storybook, you’re ready to start building your Prediction. The next tool in our Toolkit, Building Blocks, will guide you through the development steps of creating your Prediction. There are five levels of Building Blocks, each coming with a key concept you’ll need to answer; Dataset, Segment, Predicted Field, Example Set and Prediction Set.
Make sure you complete the Building Blocks in order. Let’s cover the five Building Blocks in further detail:
- Dataset: Which object does your prediction live on? Note the Dataset must come from a single Salesforce object. Tip: To grab data from Child objects, use Roll-Up Summary fields
- Segment: Should the prediction be filtered to a certain segment (sub-set) of the dataset? If so, what’s the rule criteria? This is an optional step; you don’t need to set a segment, in which case the prediction would be based on the entire dataset. Tip: set a segment if parts of your dataset are inherently different, such as Sales vs. Service Opportunities
- Predicted Field: Field that Einstein predicts. This can be a Checkbox, Numeric, or Formula Field. If there isn’t a field that can answer your prediction question, you can use filters to set up your prediction.
- Example Set: What records within the Segment should the prediction be based on? Tip: Set Example Set to records where the prediction is already known, such as Closed Opportunities (i.e. if you’re predicting Opportunity Close Likelihood, you don’t want to predict Opportunities already closed)
- Prediction Set: Records that Einstein will score. Your prediction set is automatically set as the records within Segment that don’t match Example Set criteria (i.e. if your Example Set is closed opportunities, your Prediction Set is open opportunities)
Cloudy is ready to start building their Prediction. They have completed their Building Blocks; let’s see how they look.
- Dataset: Cloudy will predict churn likelihood on the Accounts object
- Segment: Cloudy’s Accounts consist of Customers, Prospects, Partners and Vendors. Cloudy only wants to predict churn likelihood on existing customers, since the other account types do not use Cloudy’s products. Cloudy has a Boolean field on Accounts that captures whether it is a Customer, so they use that as the Segment.
- Example Set: Cloudy’s customers have contracts ranging from 12 to 36 months. Cloudy wants Einstein to base predictions on customers that have been up for a renewal recently, as these Accounts had to decide between continuing as a customer or churning. Cloudy has a Boolean field that captures whether the Customer was up for a renewal over the past 12 months, which they use for the Example Set.
- Prediction Set: Cloudy’s prediction set would be all Customers that weren’t up for a renewal recently; in other words, this would contain Accounts that are due for a renewal soon.
Completing the Building Blocks will help take Cloudy through all of the setup steps of Prediction Builder in Salesforce. Since there is great Salesforce Help documentation on that, we won’t go through that here, but if you need help, look here. After building the Prediction, Cloudy can’t enable it just yet; to truly unleash the power of a prediction, we need to empower users to take action off of the prediction. Let’s check out the Action Generator to learn how we can do that.
Tool 3: Action Generator
Now that you’ve created your prediction, you need to create automation that will let your users’ take action on the prediction. The objective of the Action Generator is to create automated actions that will help improve the business problem (which we set in the first Tool, the Prediction Storybook).
We recommend using the Prediction Scorecard (Navigation Tip: from Einstein Prediction Builder, click the dropdown next to your prediction, then Prediction Scorecard) to help with what actions to create. The Prediction Scorecard shows which fields from your prediction have the highest impact on the outcome, which are shown in the Top Predictors section; make careful note of these. Since these fields have a high impact on the outcome you’re predicting, they are a good candidate to include as a factor in your automated actions.
The Action Generator buckets predicted scores into high, medium and low, which you may set as you wish via the score thresholds. Next, set a Factor; we recommend using one of the Top Predictors from the Prediction Scorecard, as these have the power to improve the business outcome. Lastly, set an automated action; this is the action that would be kicked off in Salesforce when the predicted score falls within that threshold. This could be something like automatically creating an Event that schedules an onsite visit to the customer, or adding them to a Campaign. You may use any automation tool, like Workflows, Process Builder and Next Best Action, to build the automation. As there are likely multiple factors you’ll want to create automation off of, you may use this tool multiple times.
Cloudy checks their Prediction Scorecard, and sees that Support Level has numerous values as top predictors, so uses that factor with the Action Generator.
If the Predicted Churn is high, Cloudy wants to take the drastic action of offering free Premium Support for one year, while if it’s low, the customer would keep their current Support Level. Cloudy builds this automation in Process Builder, and filters the entry criteria based on the score thresholds and to customers that have no Support Level. Cloudy is confident that this will help drive down Churn Likelihood and ultimately save customers from leaving.
After setting up the automation, Cloudy is ready to enable the Churn Likelihood prediction and roll it out to users. But we still have one more tool in our Toolkit. Enabling the prediction will give scores on a per-record basis, but our last tool will help give insight at a higher-level view.
Tool 4: Dashboard Builder
After you’ve enabled your prediction, all of the records within your Prediction Set (which you set with Tool 2, Building Blocks) will have scores. In addition to seeing scores on a per-record basis, you probably also want to see scores at a higher-level view. This will let you see scores company-wide and drill down into records that need further attention. Our last tool, the Dashboard Builder, will help build that view. This tool works best by using Einstein Analytics, although standard dashboards also work.
The Dashboard Builder is a blank dashboard template, with different widgets and chart types you can paste into blank sections. To complete this, take a dashboard component from the widget section, fill in the blanks, and paste them into the dashboard template. When you’re done, build the dashboard in Einstein Analytics (or standard Dashboard). The dashboard template is explained in further detail by the section below:
- KPI Section: Paste 3 of your most key KPIs around the prediction
- Score Summary Section: Paste a visualization that summarizes the breakdown of your predicted scores
- Segments Section: Paste 2 visualizations broken down by a key segment for the predicted score
- Detail Table: will have the record name of the object you’re predicting on, the predicted score, and about 4 additional dimensions of your choosing as columns
Dashboard Template paste widgets into blank sections below
Widgets paste one of these widgets into the blank sections above
Here we can see how Cloudy’s Dashboard Builder looks:
Next, Cloudy creates the dashboard in Einstein Analytics, which we can see here.
This dashboard helps Cloudy drill down on focus areas quickly. For example, Cloudy has an upcoming business trip to Pennsylvania, and wants to visit his PA Accounts with the highest churn likelihood on site. All Cloudy has to do is filter the dashboard to Pennsylvania, then use the detail table to see which Accounts have the highest Churn Likelihood.
Summary
Creating a meaningful, impactful prediction can seem like a daunting task. We’ve made this Toolkit to make it easier for you. Here are all the Tools in our Toolkit, with detail on what their purpose is:
- Prediction Storybook: Help you determine what you’ll predict with Prediction Builder
- Building Blocks: Build the Prediction
- Action Generator: Create automated action off of the Prediction
- Dashboard Builder: Build a dashboard around the Prediction that gives a big-picture view of it.
Visit this link to download the Toolkit. As a reminder, make a Copy of the Slides, which you can then use to help guide you through the process of creating your first prediction!
Appendix
Here are links to other helpful tools and links on Einstein Prediction Builder
Salesforce Help Documentation — Set Up a Prediction
