About Us 
Careers Graduates
Project #1 “Column-store data warehouse” (software development)
In the area of media & marketing we provide our clients with tools to analyze consumer (survey) data and to apply insights for communication planning. Survey data has a very specific structure. A survey can contain up to 50,000 respondents and each respondent may be described by up to 50,000 variables. To access that data efficiently, we have developed a dedicated data warehousing environment that uses a custom column-oriented binary format. Our tools are developed using Microsoft Technologies. A shift from using C++ unmanaged technologies to more recent technologies such as ASP.Net, Silverlight and WPF has showed a number of bottlenecks in our data warehousing environment. To solve these bottlenecks, such as limited support for multi-user environments and the need to integrate our data warehouse in managed .Net solutions, a thorough software design is needed for a new or updated data warehousing environment. The goal of the project is to build a working prototype.
Project #2 “Consumer segmentation” (marketing / econometrics)
Pointlogic works with advertisers and agencies to optimize communication strategies and planning. The tools that we provide our clients are often based on consumer surveys. In these surveys we collect information such as:
- Demographic information (age, gender, income, region, …)
- Media consumption (does the respondent watch TV or listen to the radio, …)
- Attitudes (does the respondent like advertisement, does the respondent consider to buy a new Toyota, think Toyota is 'green')
- Important factors for buying products (is price important for the respondent when he buys a car, …)
An important factor in defining a communication plan is the target audience. A target audience may be defined with mere demographics (e.g. male between 18-34) or on other information as well (e.g. female between 18-34 who consider buying a new car). Another way of defining target audiences is by using segmentation. A segment is a classification of consumers which is based on a combination of variables. An example of a segment could be “family person”. People in this group may be assigned to this group by for example: having children, agreeing that family is important, being between 18 and 55 years old… Another example of a totally independent segment could be “new car buyers”. People are assigned to this group by for example: agree that they would like to buy a car at least every two years, having bought a car in the last two years, saying that comfort is important.
A way of defining segmentations would be by using cluster analyses. This method can be used for grouping respondents in different segments.
In this project, we will find a way to implement the segmentations for two of our main surveys. One of the surveys is about media consumption and media behaviour. The other survey is about attitudes of respondents towards buying cars. This survey has a lot of 'missing data'.
Goal of this project is to find a way to generate segments that are truly different from each other; they should evaluate cars and media differently. There are two options;
- Predefined segments
- An interactive capability that enables our clients to define their own segmentation.
Project #3 “Imputing missing values” (software development / statistics)
A lot of marketing and media research is based on consumer surveys. Unfortunately, questionnaires cannot be too long because respondents get bored and bail out or they provide ‘low quality data’. As a result, different respondents get different questions and provide for different ‘pieces of the puzzle’. What we like to do is build a tool that ‘imputes’ missing data (i.e., fill in the answers to questions that respondents did not answer). In this project we will apply methods from the statistical theory of missing values and take advantage of the structure of questionnaires to build a tool that imputes missing values. This project requires a firm statistical background, familiarity with linear regression, calculation of confidence intervals, etc.
Project #4 “Funnel Strategies” (econometrics / stochastic processes)
Many advertisers see their (potential) customers as if moving through a funnel (Dutch: 'trechter'). First, consumers have to be aware of their brand, this is stage 1. Then they have to know what the brand has to offer, and have a feeling for what it stands for. Then there is a stage of consideration, stage 3. Other examples of stages are trialists, loyal customers and advocates. Every advertiser has his own way of defining the funnel stages but the idea is the same and the goal is too: to understand what stage has the largest opportunity for the brand. Should to advertiser focus his efforts on generating more awareness among or should he focus on existing clients and try to make the more loyal? Using regular consumer surveys we can measure the size and the flow though the funnel. In this project we will assign values to each of the funnel stages and build a stochastic model to determine optimal short and long term strategies.
Careers Graduates
Opportunities for students
Project #1 “Column-store data warehouse” (software development)
In the area of media & marketing we provide our clients with tools to analyze consumer (survey) data and to apply insights for communication planning. Survey data has a very specific structure. A survey can contain up to 50,000 respondents and each respondent may be described by up to 50,000 variables. To access that data efficiently, we have developed a dedicated data warehousing environment that uses a custom column-oriented binary format. Our tools are developed using Microsoft Technologies. A shift from using C++ unmanaged technologies to more recent technologies such as ASP.Net, Silverlight and WPF has showed a number of bottlenecks in our data warehousing environment. To solve these bottlenecks, such as limited support for multi-user environments and the need to integrate our data warehouse in managed .Net solutions, a thorough software design is needed for a new or updated data warehousing environment. The goal of the project is to build a working prototype.
Project #2 “Consumer segmentation” (marketing / econometrics)
Pointlogic works with advertisers and agencies to optimize communication strategies and planning. The tools that we provide our clients are often based on consumer surveys. In these surveys we collect information such as:
- Demographic information (age, gender, income, region, …)
- Media consumption (does the respondent watch TV or listen to the radio, …)
- Attitudes (does the respondent like advertisement, does the respondent consider to buy a new Toyota, think Toyota is 'green')
- Important factors for buying products (is price important for the respondent when he buys a car, …)
An important factor in defining a communication plan is the target audience. A target audience may be defined with mere demographics (e.g. male between 18-34) or on other information as well (e.g. female between 18-34 who consider buying a new car). Another way of defining target audiences is by using segmentation. A segment is a classification of consumers which is based on a combination of variables. An example of a segment could be “family person”. People in this group may be assigned to this group by for example: having children, agreeing that family is important, being between 18 and 55 years old… Another example of a totally independent segment could be “new car buyers”. People are assigned to this group by for example: agree that they would like to buy a car at least every two years, having bought a car in the last two years, saying that comfort is important.
A way of defining segmentations would be by using cluster analyses. This method can be used for grouping respondents in different segments.
In this project, we will find a way to implement the segmentations for two of our main surveys. One of the surveys is about media consumption and media behaviour. The other survey is about attitudes of respondents towards buying cars. This survey has a lot of 'missing data'.
Goal of this project is to find a way to generate segments that are truly different from each other; they should evaluate cars and media differently. There are two options;
- Predefined segments
- An interactive capability that enables our clients to define their own segmentation.
Project #3 “Imputing missing values” (software development / statistics)
A lot of marketing and media research is based on consumer surveys. Unfortunately, questionnaires cannot be too long because respondents get bored and bail out or they provide ‘low quality data’. As a result, different respondents get different questions and provide for different ‘pieces of the puzzle’. What we like to do is build a tool that ‘imputes’ missing data (i.e., fill in the answers to questions that respondents did not answer). In this project we will apply methods from the statistical theory of missing values and take advantage of the structure of questionnaires to build a tool that imputes missing values. This project requires a firm statistical background, familiarity with linear regression, calculation of confidence intervals, etc.
Project #4 “Funnel Strategies” (econometrics / stochastic processes)
Many advertisers see their (potential) customers as if moving through a funnel (Dutch: 'trechter'). First, consumers have to be aware of their brand, this is stage 1. Then they have to know what the brand has to offer, and have a feeling for what it stands for. Then there is a stage of consideration, stage 3. Other examples of stages are trialists, loyal customers and advocates. Every advertiser has his own way of defining the funnel stages but the idea is the same and the goal is too: to understand what stage has the largest opportunity for the brand. Should to advertiser focus his efforts on generating more awareness among or should he focus on existing clients and try to make the more loyal? Using regular consumer surveys we can measure the size and the flow though the funnel. In this project we will assign values to each of the funnel stages and build a stochastic model to determine optimal short and long term strategies.