User Behavior Prediction with AI
User Behavior Prediction with AI
A marketplace website client that specializes in providing a comprehensive listing of new homes for sale across the United States. It serves as a resource for homebuyers looking to purchase a new construction home, offering detailed information on various home builders, communities, and floor plans.
Industry
Market Intelligence
Project
AI Strategy
Challenge
The client wished to track user behavior on their website to enable data-driven decision-making.
A key objective was to gauge real-time engagement levels of potential home buyers when they expressed interest in new home listings. In addition, the client wished to predict buyers’ likelihood of conversion to sales leads, by drawing analogies between their engagement patterns and those of other successful conversions.
Solutions
EX Squared delivered a robust AI scaler.
We defined the events and properties needed to capture relevant user behavior data.
The application was deployed within the client’s infrastructure using Octopus and integrated with their lead recording system for real-time predictions.
Data was collected from Twilio Segment and BigQuery (BQ) to understand the storage structure and analyze it before finalizing the desired data for prediction. Given billions of records, an efficient and cost-effective method was needed to ensure accurate predictions with a good confidence score.
We studied how various e-commerce businesses classify users for personalization. Based on our knowledge of e-commerce personalization market research, we defined a custom set of five user categories: Agents, Researchers, Price-Sensitive, Focused, and Impulse Buyers.
Using the identified traits, we developed several classification models to predict user behavior. Models were evaluated against matching home buyers, and additional scoring was done to determine more fine-grained differences among users.
Results
The AI predictive data generated by EX Squared’s behavioral model exhibited strong correlation with actual sales data when compared with other models.
Talk With Us
Tell us what you're working on. We'll ask questions, then come back with a straight view on what's worth doing, including if the answer is not yet.
Prefer to start with a call?
Book A Call