Many of you are probably familiar with the four stages of Analytics: Descriptive, Diagnostic, Predictive, and Prescriptive. If I were to ask which stages you want to achieve in Salesforce, you’d probably answer “All of them!” This blog will describe how to implement all four stages of Analytics with Salesforce Einstein in a single view. And by a single view, I mean within one record page, without the need to scroll.
To implement this view, we’ll use Einstein Analytics, Einstein Discovery, and Einstein Next Best Action.
To start, it is important to analyze the four stages of analytics: Descriptive, Diagnostic, Predictive, and Prescriptive. Let’s cover what question each stage answers:
- Descriptive: “What happened?”
- Diagnostic: “Why did it happen?”
- Predictive: “What will happen?”
- Prescriptive: “How can we make it happen?”
These stages move sequentially in difficulty and value. Let’s examine how we can use Einstein to achieve each of these stages, and walk through a real-life example of how they can be implemented in your Salesforce org.
This visual from Gartner shows the four stages of Analytics.
A key objective of Descriptive Analytics is to have insight into what’s happened at your business. Build a dashboard in Einstein Analytics to achieve this. To be able to achieve all stages of analytics in a single view, embed the dashboard on Salesforce record pages. To illustrate this, let’s put ourselves in the shoes of Cloudy’s Computing Co., a SaaS company. Cloudy needs to have a 360-degree view of their Accounts, with insight into key KPIs such as Pipeline, won and lost business, and open Cases and Activities. They’ve built an Account-360 dashboard in Einstein Analytics with these KPIs and embedded it on the Account record page. Cloudy can now answer “What happened” whenever viewing an Account. Next, let’s see how Cloudy can drill down on these KPIs, answering the “Why?” and achieving the Diagnostic stage of Analytics.
This embedded dashboard on an Account record page achieves the Descriptive stage of Analytics. We see descriptive KPIs, such as Pipeline Amount and # High-Priority Cases, and have good insight into what’s happened at the Account.
A key characteristic of Diagnostic Analytics is the ability to drill -down into data, which we can achieve via standard faceting, drop-down filters, lens exploration, and detail tables in Einstein Analytics. Let’s revisit Cloudy’s use case to illustrate this. Starting with the embedded Account-360 dashboard, Cloudy has built supplemental dashboards that drill into Opportunities, Tasks, and Cases in further detail. Cloudy can click on the Tasks link at the top of the dashboard, then use faceting to drill-down into a view Cloudy wants to gain further insight on. Cloudy can now see the specific details around the open Task assigned to Nicolai. At this stage, we’ve used an embedded Einstein Analytics dashboard to achieve the Descriptive and Diagnostic stages of Analytics. Let’s see how Cloudy can use Einstein Discovery to predict future outcomes at the Account and achieve Predictive Analytics.
With the ability to drill down on dashboards, we achieve Diagnostic Analytics. Here, we filter down to Tasks with the Status of Not Started and Owned by Nicolai Johnson. The table shows us all the detail around that Task.
A key objective of Predictive Analytics is to predict future outcomes at your business, which we can do with Einstein Discovery (or Einstein Prediction Builder, as an alternative). To achieve this, we simply build a predictive model in Einstein Discovery, then Deploy it to the object you want to predict individual records on. Putting ourselves back in the shoes of Cloudy, they’ve recently been experiencing an issue with higher customer churn (AKA attrition or turnover), but they can’t pinpoint why. So they build a predictive model in Einstein Discovery that measures the likelihood to churn across a number of key variables (industry, buyer profile, amount spent annually, service tickets, training taken, etc.) and deploy it to the Account object. When viewing an Account record, Cloudy now has insight into the Account’s churn likelihood, and visibility into the leading causes of the score and ways the churn likelihood score can be improved. Now Cloudy is set up to take action on these insights.
The Einstein Discovery Propensity to Churn card is the Predictive stage of Analytics. Here, we can see the Account’s likelihood to churn. We can also see the score’s leading causes and how it can be improved.
At this stage, we’ve achieved Descriptive and Diagnostic Analytics via an embedded Einstein Analytics dashboard, and Predictive Analytics via an Einstein Discovery model that is placed directly to the right of the embedded dashboard. Within one view, we’ve achieved the first three stages of Analytics. Let’s see how Cloudy can use Next Best Action to take preventative actions to decrease churn likelihood, and achieve the last stage of Analytics.
After adding the Einstein Discovery card to the Account layout, Cloudy has achieved the first three stages of Analytics; Descriptive and Diagnostic via the embedded Analytics Dashboard, and Predictive via the Propensity to Churn model.
A key component of Prescriptive Analytics is the ability to take action on insights. With its recommended action framework, Next Best Action is the perfect tool for this. Let’s revisit Cloudy’s use case. Cloudy bases his recommendations on the predicted propensity to churn the score of an Account. If it’s high, they recommend drastic actions, such as offering free promotions or upgrades. Cloudy builds logic to surface multiple recommendations at once and lets the salespeople decide which Recommendation to accept.
The Next Best Action card is the Prescriptive stage of Analytics. Based on the predicted Propensity to Churn from the previous step, we can take action on that via Next Best Action Recommendations. It is up to the end-user to decide which Recommendation to accept.
After adding the Next Best Action Card below the Einstein Discovery card, Cloudy has achieved every stage of Analytics. Within one view, we’ve got the embedded Einstein Account-360 Dashboard, Einstein Discovery propensity to churn model, and Einstein Next Best Action Recommendations. Cloudy navigates from gaining insight into the key KPIs at the Account via the embedded Analytics dashboard, to taking action on a Next Best Action Recommendation that can ultimately help save the customer.
This view of an Account record achieves all four stages of Analytics. The embedded Analytics dashboard achieves the Descriptive and Diagnostic stages, the Discovery Propensity to Churn model the Predictive stage, and the Next Best Action Recommendations the Prescriptive stage.
Let’s recap what we’ve done to achieve each stage of Analytics:
- Descriptive: Created a dashboard in Einstein Analytics and embedded it on a Salesforce record page.
- Diagnostic: Used drill-down capabilities, such as faceting, filters, and detail tables, on the embedded Einstein Analytics dashboard.
- Predictive: Built a predictive model in Einstein Discovery and deployed it to the same object the embedded Analytics dashboard is on. Placed the Einstein Discovery model to the right of the embedded dashboard.
- Prescriptive: Built Einstein Next Best Action Recommendations based on the predicted score from the Discovery model. Placed the Recommendations below the Einstein Discovery model.
All four of these stages are visible on the record page without the need to scroll. Navigating between all of these stages in a single view has enabled us to build an Intelligent CRM!