Einstein Discovery Desktop and Server extensions allow you to interact with the model you developed in Tableau CRM. You can do so much with that model utilizing real-time interactions, model refreshes, and brilliant insights. However, users without a Tableau CRM license will not be able to interact with the visualization. In addition, the Einstein Discovery extension cannot be accessed in Tableau Public - this makes it challenging for our community to engage with the exciting new Tableau/Salesforce feature. Aside from the two visualizations shared on our Tableau Public page, I am unaware of any other examples of Tableau CRM in Tableau Public.
There is a way to get some of these features in Tableau Public. Unfortunately, it doesn’t contain all of the potential Tableau CRM with a Desktop/Server extension offers. Still, you can apply its AI potency with your data to answer your question and interact fluidly with the rest of your visualization.
Tools Needed:
A Tableau CRM Data Model (free sandbox version)
Tableau Prep (two weeks free every quarterly release)
Tableau Desktop (two weeks free every quarterly release)
This article is intended for those that have or plan to learn how to develop a Tableau CRM model. (1)
Important: The model data from Tableau CRM and data in Tableau need to be consistent. Multiple data sources can be used in Tableau, but one should contain the fields in Tableau CRM to be appropriately linked.
Tableau Prep
Once the model is deployed in Tableau CRM, linking to Tableau becomes possible. The method shared here will not use any extensions on Tableau Desktop. However, Tableau Prep is necessary to access Tableau CRM and append it to the original data source. If you haven’t used Tableau Prep before, there is nothing to fear; since version 2021.1. Tableau makes it easy to bring in the data.
Data Source
Open Tableau Prep and access the data source which includes the Tableau CRM fields. Next, click the plus sign to the right of the step and select ‘Prediction.’ This piece engages Tableau CRM. Note: In this example, the data is already cleaned, so ‘Clean Step’ is unnecessary.
Prediction Step
Once accessed, you can begin working with the prediction step from Tableau CRM. Here are the points to complete the prediction step:
Select ‘Prediction’ to highlight.
Go to ‘Settings’ and make sure ‘Top Predictors’ and ‘Top Improvements’ are not checked. (2)
Make sure every field in Tableau CRM (Model Fields) has a corresponding field in Tableau ‘Flow Field.’
When all three points are covered, you can select ‘Apply’ to complete the ‘Prediction’ step.
New Data Source
Output
Since we incorporated the Prediction Step, we can create a new data source in Tableau.
Select ‘Output’ to edit its settings.
We want to create a file to use as a data source, so select ‘File.’
Additional Settings to update:
Add a name to the data source. In this case, it’s named ‘My New Data Source.’
Select file location.
Determine the file type. In this case, I am saving as a .csv.
Select ‘Run Flow’ to save the file.
Adding Tableau CRM Data to Tableau Desktop
Now you can apply the data from Tableau CRM in Tableau! Again, we want to answer the primary question and interact with the ‘Prediction’ field using filters. Note: This tutorial doesn’t cover building a Tableau visualization but walks you through adding the Tableau CRM context in an existing viz with a new sheet.
To make the visualization resemble how the Einstein Discovery extension works in Tableau, I attached a copy of the worksheet below.
Beneath, I included the filters and sheet actions to work with the data, how I want it calculated, and showing the impacted rows.
The field will always default to ‘Prediction.’ In most cases, you will want to get an average of the measure because the data works with each row separately. In addition, you will want to access the count of rows or records impacted as it tells the user the number of items it’s basing its prediction. Once done, simply add the sheet to the dashboard like any other sheet.
For my visualization, I wanted to add a little more information to explain this resource to the user and use a floating object to show what the extension would look like and a show/hide button with the sharp Einstein logo.
Here is the interactive visualization applying Keyrus’ technique:
Extra credit
Share your enhanced visualization in social media and tag #Keyrus and Adam Mico. We would love to see what you come up with and develop with this sensational new feature!
Footnotes:
For those who haven’t yet, but are interested, please check out these free Trailhead training paths:
Note: ‘Top Predictors’ and ‘Top Improvements’ are helpful with the Einstein Discovery extension because it interacts with the model, but less effective if using as a static data source. In this use case, we want to interact with the primary question asked.