About the Growth Play

Drive more revenue with tailored upsells leverage customer data to predict and recommend product categories most likely to lead to additional purchases. By leveraging the ML-based lifecycle agent, you can create journeys that message users via email and SMS with categories that increase the likelihood of purchasing. You can also adjust your website experience to provide personalized banners with suggestions for users to explore the recommended category based on their real-time browsing behavior.

‍

Benefits

• Increased average order value (AOV). Customers are more likely to add more items to their cart by recommending additional relevant product categories.
• Enhanced customer experience. Personalized recommendations make shopping more enjoyable and relevant for customers.
• Higher conversion rates. Targeted upsells are more likely to convert than generic promotions.
• Improved customer retention. Satisfied customers are more likely to return, boosting long-term loyalty.
• Operational efficiency. Automated recommendations reduce the need for manual intervention, saving time and resources.

‍

How it works

Step 1: Define and create events for each category

1. Create category-specific events:

• Identify the product categories you want to track (e.g., dresses, shoes, accessories, outerwear).
• For each category, create an event. For example:
  • "Visited dresses category"
  • "Visited accessories category"
  • Visited outerwear category"
• Use the event type "Page view" and include the category's URL parameters to track these events accurately. For instance, track page views with URLs containing /category/dresses, /category/shoes, etc.

‍

Step 2: Build the lifecycle agent

1. Set the goal event:

• Define "Purchase" as the prediction agent's goal event. This can be tracked by setting up a "Purchase" event that triggers when a transaction is completed.

2. Add category events to the agent:

• Include all the category-specific events created in Step 1 as potential actions for the model to choose from. Each category event should be associated with the goal event "Purchase".

‍

3. Filter training data:

• To ensure the agent is trained on relevant data, exclude users who do not fit your desired criteria. For example, exclude users who have not purchased in the last year.

4. Train the agent:

• Allow the agent to analyze past data to identify patterns and predict which category visit will most likely lead to a purchase. This involves feeding historical data of customer interactions and purchase history into the model.

5. Agent output:

• The model will output an attribute with a value reflecting the category that has the highest likelihood of leading to a purchase for each customer. For instance, if the model predicts a high likelihood for "Visited shoes," it will output an attribute like shoes_category.

‍

Step 3: Create segments based on model output

1. Create a new segment:

• Navigate to the Segments section in Intempt and create a new segment.

2. Define segment conditions:

• Use the attribute generated by the next best action model to define the segment. For example:
• Segment: Users with the attribute value shoes_category.
• Condition: Attribute next_best_action equals shoes_category.

3. Examples of configuring segments:

• High likelihood to purchase dresses
• Segment name: "High likelihood to purchase dresses"
• Condition: Attribute next_best_action equals dresses_category

High likelihood to purchase accessories:

• Segment name: "High likelihood to purchase accessories"
• Condition: Attribute next_best_action equals accessories_category

High likelihood to purchase outerwear:

• Segment name: "High likelihood to purchase outerwear"
• Condition: Attribute next_best_action equals outerwear_category

4. Combine with other conditions:

• You can further refine these segments by adding additional conditions. For example:
• High likelihood to purchase dresses and recent activity:
• Segment name: "High likelihood to purchase dresses and active in last 30 days"
• Conditions:  
  • Attribute next_best_action equals dresses_category
  • Last activity date is within the last 30 days
• High likelihood to purchase shoes with high average order value:
• Segment name: "High likelihood to purchase shoes and high AOV"
• Conditions:  
  • Attribute next_best_action equals shoes_category
  • Average order value greater than $100

‍

Step 4: Set up personalized journeys

1. Create a new journey:

• Navigate to the Journeys section in Intempt and create a new journey.

2. Define the journey trigger:

• Set the trigger for the journey based on the segment created in Step 3. For example:
• Trigger: When a user enters the "High likelihood to purchase shoes" segment.

‍

3. Add actions to the journey:

• Example nurture flow for High Likelihood to Purchase Shoes:
• Step 1: Initial Email:  
  • Action: Send a personalized email featuring the shoes category.
  • Example:Subject: Discover Our Latest Shoe Collection!
  • Content: Hi [Name],We noticed you are keen on our shoe collection. Check out our latest arrivals and find the perfect pair to complete your look.[Shop Now Link]Best,The Thread.ly Team

‍

‍

• Step 2: Follow-up Email:  
  • Action: Send a follow-up email with a special offer.
  • Example: Subject: Special Offer on Shoes Just for You!
  • Content: Dear [Name],We’re offering you an exclusive discount on our shoe collection to show our appreciation. Don’t miss out on this limited-time offer![Get Your Discount Link]Happy Shopping,The Thread.ly Team

‍

• Step 3: Reminder Email:  
  • Action: Send a reminder email if no purchase is made.
  • Example: Subject: Last Chance to Save on Your Favorite Shoes!
  • Content: Hello [Name],This is your last chance to take advantage of our exclusive discount on shoes. Hurry, the offer ends soon![Shop Now Link]Best,The Thread.ly Team

‍

4. Add decision points:

• Incorporate decision points to refine the journey further. For example:
• Decision: If the user clicks on the email link, proceed to show a special discount offer.
• Decision: If the user does not engage with the email, send a follow-up reminder after a few days.

‍

5. Activate the journey:

• Review the journey steps and activate it to start the automation process.

6. Multiple journeys for different categories:

• For each category, create a separate journey trigger and associated actions. For example:
• If you have 30 categories, set up 30 journey triggers and corresponding actions.

‍

Step 5: Deploy personalizations

1. Create a new personalization campaign:

• Navigate to the Personalizations section in Intempt and create a new campaign.

2. Create multiple experiences within the campaign:

• For each category segment, create a separate experience within the personalization campaign.

3. Set up the personalization content for each experience:

• Experience for Shoes Category:  
  • Homepage Banner: Display a banner featuring the latest shoes collection.
  • Sticky Top Bar: "Check out our newest shoe arrivals, [Name]!"
  • Recommended Product Section: Recommend popular and new shoe arrivals.
  • Offer section: "Exclusive Offer: 10% off all shoes for a limited time!"

• Experience for Dresses Category:
  • Homepage Banner: Display a banner featuring the latest dresses collection.
  • Sticky Top Bar: "Discover our stunning new dresses, [Name]!"
  • Recommended Product Section: Recommend popular and new dress arrivals.
  • Offer section: "Special discount: 15% off on new dresses!"

• Experience for Accessories Category:
  • Homepage Banner: Display a banner featuring the latest accessories collection.
  • Sticky Top Bar: "Complete your look with our new accessories, [Name]!"
  • Recommended Product Section: Recommend popular and new accessory arrivals.
  • Offer section: "Limited Time Offer: 20% off all accessories!"

‍

4. Set the targeting conditions for each experience:

• Define rules to ensure the personalized content is shown to the right users. For example:
• Experience for Shoes Category: Target users where the attribute next_best_action equals shoes_category.
• Experience for Dresses Category: Target users where the attribute next_best_action equals dresses_category.
• Experience for Accessories Category: Target users where the attribute next_best_action equals accessories_category.

‍

5. Activate the personalization campaign:

• Review the personalization setup and activate the campaign to start delivering personalized experiences.

‍

Step 6: Monitor and optimize

1. Track performance with journey analytics:

• Navigate to the Journey Analytics section in Intempt to analyze the performance of your journeys.
• Key metrics to monitor:
  • Triggered journey: Number of users who have triggered the journey.
  • Converted: Number of users who completed the conversion event set for the journey.
  • Conversion rate: Percentage of users who converted out of those who triggered the journey.
  • Days to convert (Avg): Average number of days it took for users to convert after triggering the journey.
• Use this metrics to evaluate initial engagement, optimize timing, and identify areas for improvement. For example, if the conversion rate is lower than expected, consider adjusting your emails' and SMS messages' content or timing.

2. Track performance with personalization analytics:

• Navigate to the Personalization analytics section in Intempt to analyze the performance of your personalization campaigns.
• Key metrics to monitor:
  • Control group: Represents the group of subjects that are set aside and do not receive the modified content.
  • Unique views: Measures the number of customers that viewed the personalized experience.
  • Conversion: Specifies the number of users that triggered the conversion metric.
  • Conversion %: Percentage of users across the audience that triggered the conversion metric.
  • Conversion value: Displayed if the conversion event includes the value of an event property.
  • Lift: Shows the improvement in the conversion rate percentage for the variants compared to the control group.

• Use this metrics to assess the effectiveness of your personalized content. For instance, if the lift is significant, it indicates that the personalized content is performing well compared to the control group. Conversely, if the conversion rate is low, you may need to refine your personalization strategy or content.

3. Refine the model:

• Regularly update and refine the next best action model based on new data to improve its accuracy and effectiveness.
• Adjust the categories and events tracked to ensure the recommendations remain relevant and impactful.

‍