Using Predictive Segmentation to Target the Right Audience That Drives Results

Demo Series Transcript: Using Predictive Segmentation to Target the Right Audience That Drives Results

Speaker:

• Stan Szeto, Senior Solution Architect, Blueshift

How Blueshift's AI Helps Predict Customer Behavior and Personalize Campaigns

Stan Szeto: Hi, I’m Stan Szeto, Senior Solution Architect at Blueshift. Thanks for joining our bi-monthly demo series. Today, I’ll walk you through how Blueshift’s AI and machine learning help predict customer intent, improve targeting, and drive better conversions.

We'll start with a quick five-minute overview of Blueshift for those new to the platform, followed by a deep dive demo into our predictive modeling and activation features. Finally, we’ll close with Q&A.

What Is Blueshift and What Makes It a Smart Customer Engagement Platform

Blueshift is a customer engagement platform built around a SmartHub CDP, enabling marketers to unify, analyze, and activate data from across all their customer touchpoints.

• It connects data from websites, mobile apps, backend systems, and point-of-sale systems to create a unified customer profile.
• It uses AI to enrich that data—powering predictions like purchase intent or churn risk.
• And it enables omnichannel orchestration across email, SMS, push, direct mail, paid media, and more.

We support enterprise-grade compliance including GDPR, CCPA, HIPAA, and SOC2.

How Predictive Scoring Works in Blueshift

Blueshift’s Predictive Studio helps marketers create AI-powered models tailored to their business needs—no data science team required. These scores can:

• Predict likelihood to purchase
• Detect churn risk
• Identify the best marketing channel for each user

Unlike black-box systems, Blueshift’s models are transparent, allowing marketers to:

• See top signals driving each score
• Analyze conversion likelihood by score band
• Continuously refine inputs based on new data

Real-World Example: A Financial Brand Predicts Product Interest

A finance client used Blueshift to build scores across multiple product lines—credit cards, auto loans, and refinancing:

• Each customer was scored on their likelihood to convert for each product
• Segments were created based on score thresholds
• Audiences were synced to paid media platforms (e.g., Pinterest, Google Ads)
• Personalized ads were served based on the highest intent score

Once users returned to the site and became known, the same scores powered consistent engagement across email, mobile, and web.

Live Demo: Personalization in Action on BluBluLemon.com

Stan walked through a live example on BluBluLemon.com:

• Viewed and added products to cart
• Removed an item

In Blueshift, the customer profile for "Stan" was instantly updated with:

• Viewed items
• Carted items
• Engagement scores across channels (email, push, SMS)
• Predictive scores (purchase intent, retention likelihood)

How to Build and Customize Predictive Scores in Blueshift

In Predictive Studio, marketers can:

• Select the goal event (e.g., purchase)
• Define funnel events leading to that goal
• Choose time windows and attribute inputs
• Launch the model with 3 steps

Models auto-train using behavioral and demographic data, and update daily. Feature importance shows top signals, like recency or frequency of site visits.

Using Predictive Scores to Build Smart Segments

Stan showed two segment examples:

High Intent Shoppers:

• Lifetime revenue > $300
• Lifetime orders > 3
• Purchase score between 75 and 100

These users were synced to paid channels like Facebook, Google, and Criteo for immediate remarketing.

At-Risk Customers:

• No site visits in 7 days
• No purchases in 30 days
• No email engagement in 2 weeks
• Retention score below 50

This segment entered a win-back flow.

How to Orchestrate Omnichannel Journeys with Predictive Scores

Stan demoed a win-back campaign using:

• A/B testing for offers (e.g., 10% vs. 15% discount)
• Conditional logic (e.g., only email users who didn’t open last message)
• Channel scoring to route users to SMS, push, or in-app
• Purchase intent scores to tailor discounts (higher for low-intent users)
• Direct mail fallback if digital channels failed

Customer Results Using Predictive Intelligence

• CarParts: Used category-level predictive scores, resulting in a 400% lift in engagement
• PayPal: Improved gross sales per session by 125% through mobile engagement scoring
• LendingTree: Used intent scores to segment users by product line, increasing revenue 35% through paid media
• Groupon: Drove higher revenue by targeting users based on product category predictions

Final Thoughts and Next Steps

Blueshift’s predictive modeling is:

• Customizable by marketers
• Continuously refined
• Ready to use across segmentation, personalization, and campaigns

For a personalized walkthrough, visit blueshift.com or reach out to your CSM.