How is data science shaping the future of audience measurement?

With the increasing need to deliver a total view of audiences across devices, we’ve seen the rise of the data scientist. These research experts are able to solve challenging research problems and make sense of the vast amounts of information available today. But what do data scientists actually do? You could be forgiven for thinking they are simply number crunchers developing sophisticated algorithms. In fact their role is so much more!

At Kantar Media we’re proud to have recruited some of the best minds in data science and research to help shape the way audiences are measured. Among them are Sushmita and David.

Watch the video below to hear Sushmita summarise what data science is, then read on for the full interview.

(Play Video)VimeoId-172769598

How did you become a data scientist?

David – I studied Telecommunications Engineering and Masters in Biomedical Engineering in Spain. Both Bachelors and Masters projects involved working with 2D and 3D images. At the time I didn’t know these were the credentials for a data scientist even though I was playing with terabytes of data! Later I moved to London to find a job and because I love working with maths and data I looked for a role as a data analyst, and joined Kantar Media.

Sushmita – I’ve always loved solving mathematical problems. Both my degrees were in Mathematics and my PhD was in statistics. I was looking for a job involving maths and statistics with a research component. The opening at Kantar Media was perfect – and so was the interview process! I was given a few days to work on a mathematical solution for a research problem. At the end of the interview process, I was offered a job as a data scientist!

Is mathematics the most important skill you need to be a data scientist?

Sushmita and David – No!

David – You can have all the mathematical knowledge in the world but without analytical ability you’re lost! You must love solving problems! I’ve been solving crosswords and puzzles since I was a child. I really like to find a problem no one else has solved, I love a challenge. On top of that you need to be able to prove your solution and explain it to people in a way they can understand irrespective of their background.

Sushmita – Of course mathematical knowledge and problem solving skills are important but the list doesn’t end there. The drive to constantly improve and optimise the solution is critical, as is being able to communicate effectively. Some of the algorithms we use for hybrid approaches to measurement are extremely complex. We need to be able to maintain the complexity – but ensure it’s understood!

That neatly leads us on to the question - what is a hybrid approach to measurement?

Sushmita – A hybrid solution is one that combines different data sources together, preserving the strengths of each dataset. We use a hybrid approach for the calibration of online viewing in the panel and the census data. Our research teams are continually designing new methodologies, and our hybrid modelling has been developed because we understood that the definition of TV was changing with the many new ways available to consume TV or video, across multiple platforms and devices. We needed to be able to offer a cross-media view of the content audiences are consuming.

David – Although many have considered hybrid measurement to mean bringing together offline (TV set) and online viewing, we define it in a far broader sense. Picking up on Sushmita’s point, data coming from a panel using PeopleMeter technology combined with measurement through the return path (RPD) or set top box is also a hybrid approach. The way we bring them together is highly dependent on the problem and data available.

What do your current roles involve?

David – We’re a growing team in Kantar Media, operating a central hub in London and with researchers in many markets across the 50+ we operate. It’s an enriching environment as we have over 10 nationalities in the London hub alone!

On a daily basis I lead our work in the measurement of refining and designing research techniques for measurement through the return path. In addition I work a lot with conducting quality control analyses across a number of TV measurement services where we’re using this measurement technique across the world.

Sushmita – I lead our work primarily in data integration projects. We’re committed to continually developing and refining our work, with many projects running for 4 years or more with multiple components including; working on the overall methodology, presenting our approach to clients, developing each algorithm, checking the results and continually working with our clients to assess how services are operating and evolving..

What’s the most important breakthrough in data science in recent years?

Sushmita – For audience measurement, the daily delivery of calibrated respondent level data as part of SKO’s Total Video Integration model has to be it for me! We designed algorithms and validated the output and together with SKO, we recently won awards at the I-COM Data Creativity Awards and the AMMA awards for this service.

David - We’ve finally broken down the wall! The wider industry now realises that we are able to measure different ways of viewing and can bring those data sets together. We’re excited to continue solving the complex research requirements of our clients, and to continue making breakthrough’s in this field.   

What are you proudest of when it comes to your work?

David – The problem I was given to solve at my interview became my first project when I started in my new role. It’s exhilarating to see that your ideas can be put into practice and become a valuable solution!

Sushmita – Some of my proudest moments are when no one knows you’ve achieved something, such as overcoming problems along the way. One such moment was in mid-2015 when I got my hands on the first ever calibrated data for internal validation. The process of developing the calibration engine involved building several algorithms, their implementation and optimisation. When the system was finally built and actual data was processed through the system, we knew we had successfully developed a solution that the industry will benefit from. This was a big achievement and a proud moment for our company.

For more information on some of the projects that Sushmita and David have worked on, click the below links:

Search article


*Fields marked with an asterisk are required