Target Optimizer, a proprietary algorithm from Numberly, identifies individuals' level of appetence for a specific product or habit in three phases: categorization, learning and prediction. It classifies individuals into five appetite levels, updated regularly, and enables effective targeting via Queryly. This system optimizes communication by reducing targets while maintaining their impact.
Introduction
Target Optimizer identifies for each individual a level of appetence for a specific product or habit.
The result of this algorithm can be used within Queryly to target only those individuals who are most appealing to your communication.
This Numberly proprietary algorithm analyzes the history of individuals on the perimeter of your data in order to determine a level of appetence.
Getting started with Target Optimizer
The algorithm is divided into 3 phases:
- Categorization: Target product or habit selection
- Learning: From past actions, Target Optimizer learns the patterns that lead to buying this product or performing this action.
- Prediction: Target Optimizer calculates for each individual in the database his appetite to perform this purchase or action
Categorization
Target Optimizer will be able to identify a level of appetence in different categories:
- Travel: Target optimizer will identify travel appetencies to a destination, or to specific dates or to the purchase of an option.
Ex: Identify the level of travel appetency in January and February
Understanding travel appetencies - Order: Target optimizer allows you to identify appetites for making a purchase of an item or category of item, or for shopping in a specific boutique.
Ex: Identify the level of appetite for a purchase of a dress
Understanding ordering appetites - Contract: Target Optimizer identifies appetencies to take out a contract
Ex: Identify the level of appetency to take out a home loan
Understanding contract appetencies
In order to set up the appetence calculation on the desired category, a request to support will enable its implementation later this week.
The request must contain:
- The desired category: such as Travel - "Departure Month" or Order - "Category"
- The detail for the given category: For departure month: "January and February" or for category: "Skirt"
Learning
During the learning phase, Target Optimizer will go through all the data in your database to identify the patterns of actions and information that lead to the desired action or purchase.
These patterns are updated every 3 months to ensure that the model evolves in line with your customers' changing habits.
To ensure optimal training, we recommend that the database contain at least 50,000 records and at least 1,000 purchases or actions to train on.
Prediction
Looking at these patterns, Target Optimizer will calculate for each individual its probability of carrying out the action or purchase in question.
The AI will then group individuals on 5 levels of appetence:
- Maximum Appetence (5)
- High appetence (4)
- Moderate appetence (3)
- Low appetence (2)
- No appetence (1)
This appetence score is updated weekly to ensure that changes or new arrivals are quickly taken into account.
Using the result
The target Optimizer result is then available in Queryly in the Score category.
Two fields are required:
- The score field for indicating which score to analyze
- The value field for indicating the appetence level(s) of individuals to be included in the target
Conclusion
Target Optimizer enables you to identify signs of appetence in individuals and thus narrow your targets while maintaining the impact of your communications.
In order to set up a score, this guide tells you how to make this request: Request target Optimizer score set-up
.