Retail
❯
Purchase Prognosis
❯Retail - Purchase Prognosis
Large Companies
Small Companies
Retail - Purchase Prognosis
For:
Digital marketersGoal:
Improve Operation Efficiency, Increase Revenues, Reduce costsProblem addressed
Anticipating purchase demand is crucial for planning operational activities, such as logistics and inventory management in E-commerce stores. Demand forecasting based on statistical and machine learning methods using historical transaction records as input is commonplace.
Predicting intent during an online session is much less commonplace and made much harder by the fact that the majority of web browsing is still carried out anonymously. Anonymous in-session purchase intent prediction can increase conversion by serving additional tailored content to the end user and protect margin by suppressing unnecessary discount offers being shown.
Scope of use case
Increasing revenue per user on an E-commerce site through in-session purchase intent prediction.
Description
Machine learning models can be trained on the session logs of ecommerce websites to score the likelihood of a user session ending in a purchase. Browsing patterns, session duration and temporal features, amongst many other factors provide important indicators of purchase intent. Combining these factors further with other inputs from real time streaming data that provide in session behaviors improves prediction accuracy. By testing model results using control groups consisting of held out web traffic, significant uplift in conversion and other metrics can be achieved.
Using ZineOne’s In-Session Marketing platform a leading US retailer was able to segment website visitors in real time into three groups:
1. those who will not buy;
2. those will buy; and
3. undecided.
By sending tailored, personalized content to each group of visitors, the retailer was able to achieve a 20% average lift in revenue per visitor.
Raw Data
AI: Perceive
Machine Learning
AI: Understand
Plan & Schedule
Predict / Forecast
AI: Act