Salesforce users live hectic lives. They have countless calls to make, tasks to complete, and events to attend. With having so many actions to get done each day, Salesforce admins worry users won’t take action on the right ones. Wouldn’t it be great if there was a tool that could recommend the next best step for users to take?
The new tool that makes this possilbe is Einstein Next Best Action. It can deliver optimal recommendations at the point of maximum impact to users. If you don’t know where to get started, keep reading! This blog post will walk you through the steps of creating your first Next Best Action Recommendation, and show you how it can guide your users to better decision making.
First, let’s review what Einstein’s Next Best Action is and when to use it. Einstein Next Best Action is a tool that surfaces recommendations to users. It’s important to distinguish that a recommendation is not full automation. Ultimately, the user decides whether or not to follow through with the recommendation. You can use Next Best Action when you want to surface recommendations to users based on predefined criteria, but still want the final decision on the next actions to lie with the user.
Before we dive into development steps, let’s review some pre-deployment steps. These are the questions you need to answer before you create your Recommendation. There is a logical progression to these questions, and we’ll provide sample answers to each.
- What am I solving?
a. ‘What cross-sell opportunities to offer?’
b. ‘Which level of support to offer?’
- What recommendations to make?
a. ‘Cross-sell Product X’
b. ‘Send Service Technician Onsite’
- Under what conditions do recommendations apply?
a. Cross-sell Product X when Opportunity also has Product Y
b. Send Service Technician onsite when product is under warranty but still repairable
Ultimately, the conditions selected in question 3 will decide when to surface recommendations to users. If the user chooses to accept the recommendation under that condition, the automation in question 2 will be kicked off.
Now that we have gone over the pre-deployment steps, let’s dive into the development steps. Creating an Einstein Next Best Action Recommendation is a 4-step process. As we walk through these 4 steps, let’s imagine we are a regional appliances company that wants to use Next Best Action to guide customer service reps to make better decisions when handling cases with customers.
1) Lightning Flow
The first step is to create a Lightning Flow that serves as the automation if the recommendation is ultimately accepted by the user. To create a Flow, navigate to Setup, search for Flows and click New Flow, then select Screen Flow. This will take you to the Flow Builder. We won’t cover the steps of creating a Flow in this blog, as that isn’t unique to Next Best Action, but you can refer to this documentation for help. In this example, if the user decides to accept the recommendation of sending a Service Technician onsite, the status of the case will be updated and an Event will be created that lets a field-service rep know they have to visit a customer to try to fix a broken product.
The next step is to create a Recommendation. To Create a Recommendation, which is a standard Salesforce object, navigate to the App Selector and select Recommendations, then New. The Recommendation appears to the user via the UI on whatever object it applies, which in our example is Case. Here, we generate a Recommendation to dispatch a Service Technician. A key piece is that the action in the bottom right ties to the Lightning Flow that we created in the previous step. If a user accepted this recommendation, it would launch the flow from the previous step, which updates the Status of the Case record and create an Event.
3) Action Strategy
a) The third step is to create your Action Strategy. To create an Action Strategy, navigate to setup, search for and click on Next Best Action. Then, click New Strategy, give it a Name and select which object it applies to. This will bring you to the Strategy Builder, which is made up of a toolbox of various elements which funnels Recommendation records through your logic to determine Recommendations that are surfaced to users. For this blog, we’ll cover two of the most common elements, Load and Filter.
b) The Load Element will load Recommendations created in step 2. For this example, let’s use the Dispatch Service Technician Recommendation we created in the previous step. In the Load Recommendations When section, tie it to the Id of the Recommendation record.
c) The Filter Element will filter Recommendations based on criteria. For example, we may want to filter the Dispatch Service Technician Recommendation to the surface only when the Product is under warranty and repairable, which is identified via Boolean custom fields on Case. Next, connect the Load and Filter elements, so that the Service Technician recommendation is only surfaced when the criteria from the Filter element is matched. For further detail on what all of the Elements do, refer to the Salesforce documentation.
4. Display Recommendations
a. The fourth and final step is to display Recommendations on record pages. Using our previous example, we want to show the Recommendation on the Case Lightning page. Navigate to a Case record and click on the Edit Page via setup and drag an Einstein Next Best Action Lightning Component onto the page. Next, name the Next Best Action card and tie it to the Action Strategy you created in the previous step. You have now surfaced Recommendations to users!
Art of the Possible
Now that we have taken you through the journey of creating your first Next Best Action, let’s take a sneak peek at the art of the possible with NBA. Next Best Action is already a very strong tool on its own, but it can be even more powerful when used in conjunction with other Einstein tools.
When used with Einstein Discovery, we can gain AI-powered Recommendations. We can do this by displaying an Einstein Discovery prediction in a Salesforce object; documentation on this can be found here.
Let’s imagine we created a prediction in Discovery that calculates an Accounts Propensity to Churn. We display the prediction on the Account page and store the prediction value in a custom field. We now have the power to build an NBA Strategy based on the prediction value. For example, if the Propensity to Churn prediction is high, we want to nurture the Account, creating Recommendations that offer things like free support packages or in-person user training. We’ve now used an AI-powered prediction to build an even smarter Next Best Action Strategy! SpringML’s Predictive Churn Accelerator solution provides actionable insights to predict customer churn. You can get a 360-degree view of your customer to determine the purchase history giving you the ability to design marketing campaigns and plan retention programs that have maximize their impact.