Case Studies Predictive Analytics Demographic Ad Targeting

Project Objective

To develop a statistical model to estimate the demographic profiles of websites for enabling accurate audience targeting.

Client
Pioneer of online ad targeting specializing in providing innovative performance-based online media solutions for response-driven marketers, advertisers and publishers.

Approach
The modeling process was executed in three phases:

1. Identifying a stable sample of users who exhibit sufficient and regular activity. These users and their web traversal data will be used for estimating the demographic profiles of websites.
2. Demographic profiling of the users belonging to the stable sample using a set of skewed websites (both age and gender) using a secondary data source.
3. Demographic profiling of websites visited by a statistically sufficient number of profiled users at day of the week and day-part granularity.

Approach for Developing Demography Identification Model

Solution
The model calculated the skewness of traffic visiting a particular website for a particular day and day-part combination.
Select Months
Feb - 05 Mar - 05 Apr - 05 May - 05 June - 05
July - 05
Aug - 05
Sep - 05
Oct - 05
Select Days
Weekdays Weekends Mon Tue Wed Thu Fri Sat Sun
  MALE Age Groups FEMALE Age Groups
Day Parts <15 15-25 26-40 40+ <15 15-25 26-40 40+
1:00- 2:00                
2:00- 3:00                
3:00- 4:00                
                 
22:00-23:00                
23:00-24:00                                  
Sample Demographic Model Output for a Website
The model output was used to answer questions like:

1. What is the age and gender profile of a particular website on a particular time of the day.
2. A client wants to buy one million impressions to be targeted at 'Males between 26 to 40 years of age' over next 60 days. What websites will be the most appropriate?
3. Given the following requirement of audiences for next three months, what are the best sites to buy?
4. Given our existing inventory of media bought, what kind of surplus audience we have on hand?

The model output was accessed by the client's ad-serving engine for targeting audiences. Web history data for approximately 1 million users was processed on a daily basis using a scalable architecture.

Benefits
The client was able to improve the effectiveness of online ad campaigns by ensuring that online ads were placed on appropriate websites based on target audience.
 
 
Targeting Internet users by their demography based online behavior
Pioneer in online advertising solutions
Improved effectiveness of online ad campaigns by ensuring high visibility among target demographic audiences
- Behavioral Online Ad Targeting
- Churn Prediction
 
Home   |  Portfolio   |  Publications   |  Careers   |  Contact Us
© 2000-2011 DecisionCraft Inc. All Rights Reserved.