You can apply Treasure Data’s machine-learning feature, Predictive Scoring, to your customer segment. Use Predictive Scoring to assess behaviors and predict future behavior.
Leveraging machine-learning techniques is crucial to efficiently and effectively understand customer data. However, marketers who use Audience Suite do not need to be familiar with machine-learning and data science. Predictive scoring in the Audience Suite enables users to enjoy machine-learning capability in their day-to-day activities with no technical and theoretical expertise. Marketers can predict customer behavior such as who is likely to churn, purchase, click, or convert in near future.
What you need to prepare
Our predictive scoring calls for you to describe what you want in the form of batch segments:
- Population is a set of customers. The set is used to build the predictive model; the characteristics of customers are extracted and customer patterns are learned, and the predictions is based on the patterns learnt from the segment.
- Positive samples indicate your definition of “conversion”; for example, if you want a customer churn prediction, then this segment is a set of customers who were already churned.
- Scoring target specifies customers you are interested in; the prediction is done only for the customers in this segment.
Depending on configuration of the segments, you can solve a wide variety of “wants” that predict “promising” customers. As long as a subset of customers can be represented as a batch segment in an audience, Treasure Data can predict customers who are likely (or, if desired, unlikely) to be in the segment.
As a consequence of scoring, you can visually check the distribution of scores and corresponding metrics on dashboard. Additionally, syndication for “likely” customers can be seamlessly implemented from the dashboard:
This section introduces possible use cases which can be implemented using the Predictive Scoring feature in Audience Suite. You can solve any of these customer prediction “wants” on the platform by just clicking some buttons.
Predict future customer churn
- Population: All customers in an audience
- Positive samples: Already churned customers
- Scoring target: Not churned customers
Predict future conversion in US
- Population: People in US
- Positive samples: Converted customers
- Scoring target: Not converted yet customers in US
Predict customers' purchase in 2018 based on 2017’s history
- Population: People who accessed in 2017
- Positive samples: Purchase log in 2017
- Scoring target: People who accessed in 2018
Learn more about predictive customer scoring at:
- Step-by-Step Tutorial toward First Predictive Scoring
- How Feature Guess Works
- How to Tune Predictive Scoring