Next Best Action

Sales and service teams often struggle to identify the optimal follow-up steps, maintain a consistent approach, make data-driven decisions, get real-time advice, and predict high-value actions. The Next Best Action feature provides real-time, AI-powered guidance to ensure process consistency, improve conversion rates, offer personalized recommendations, automate workflows, and enhance customer engagement.

Pain points

Sales and service teams face the following challenges:
• Difficulty identifying the best next action for each customer interaction
• Inconsistent sales and service approaches across team members
• Manual decisions based on intuition rather than data
• Lack of intelligent recommendations based on real-time context
• Difficulty predicting which actions will yield the highest value

Benefits

The Next Best Action feature delivers the following benefits:
• Recommends the single, highest-value next action in real time, based on business rules and AI predictions.
• Ensures a consistent approach for all sales representatives and service staff.
• Improves conversion rates with data-driven action recommendations.
• Provides personalized recommendations based on customer history, behavior, and preferences.
• Automatically triggers workflows when the AI confidence reaches a preset threshold.
• Enhances customer engagement through timely and relevant actions.

How to use

Template type	Object reference	Data source
Flex	Account, Orders, Order Items	CRM, Flow

Create a new template based on the template shown here, and add the data you need.
For Flex templates, you can deploy them to a Lightning page or a Flow.

Prompt template
You are a sales representative for {!$Organization.Name}.
Analyze the previous order records for {!$Input:Account.Name} to determine the best recommendation for their next order.
Query the customer's order records from the past year.
{!$Flow:GenAI_Get_Last_Year_Ordered_Products.Prompt}
Summarize the products the customer purchased and count how many times each product was purchased.
Determine the last purchase date for each product.
Based on the information above, recommend 10 products for the customer's next order. Prioritize products that are purchased frequently or have not been purchased in a long time.
Also, recommend products that better fit the customer’s industry ({!$Input:Account.Industry}).
Sort the recommendations from highest to lowest priority.
For each recommended product, state how long it has been since the last purchase (in months or weeks) and clearly state how many times it was purchased in the past year.
Do not recommend items like shipping fees, as these costs are calculated automatically.