The Big Data Expo 2015: RTL & Bol.com

9 October 2015

RTL Netherlands: Media data intelligence

Geoffrey van de Meer (Manager Consumer Intelligence) told a packed room what he has achieved in the past year and a half in the area of data platforms, data collection and data science for RTL Netherlands. In addition, he showed a series of examples in whicBig Data Expoh data is used to achieve various objectives, such as:

• Price elasticity calculation: what price do we need to ask to maximise the revenue from the advertising blocks?
• Segmentation of viewers and videos. In which segment is there a risk of a shortage of new content?
• Face recognition and text-mining to gather more data about the content to connect it subsequently to customer behaviour: which people/actors provide the highest ratings, which combinations are popular and how long can someone keep talking on a talk show before viewers switch channels?
• Predicting the age of customers/website visitors to make the adverts more relevant, among other things: visitors with a browser that is not up-to-date are more often female than male, customers with an “old” email address (e.g. Hetnet, Planet, Tiscali) are often 45+.

So these are great examples of what is possible by storing data properly and applying it.

Bol.com: Every customer shops in their personal store

Yory Wollerich (Senior Customer Insights Analyst / Data Scientist) showed us how Bol.com personalises its store to increase conversion. He showed how you can generalise market research into purchasing behaviour among a limited number of customers with other visitors of the website and can use this to personalise the store and emails based on (among other things) Cialdini’s principles of persuasion. So someone who is likely to be driven by ‘social proof’ looks at the product reviews first, then the specifications. It was also indicated that Bol.com wants to respond more to major events in someone’s life, such as moving house or pregnancy. Thus for each visitor/customer it is calculated what the probability is of a house move or pregnancy. When the calculated probability is high enough then a confirmation is requested, after which the customer will be served related products (and orientation content).

To automatically respond to developments around us an algorithm is used that recommends trending products. If during the show “The world keeps turning” a book is recommended, there is a greater chance that you will immediately find the book on the homepage. Thus, even for visitors about whom little is known, the displayed product offering becomes more relevant.

Tips for your next visit

If you want to visit the Big Data Expo next year, we have a few tips for you. Be careful which talks you visit: There is a big chance that someone is speaking who does not have (hands-on) experience in converting data into insights and applications that have yielded demonstrable value to the customer and the company. Therefore, read the description beforehand in order to avoid general lectures on ‘market trends’ and pick out lectures with detailed case studies (such as those of RTL and Bol.com).

Exhibitors are especially interesting if you are responsible for the IT infrastructure and/or BI (dashboards) within a company. For other visitors: avoid companies that claim to have ‘Big Data Analytics’ and related insights and applications in a ‘product’. Usually it takes more than a good IT infrastructure for an organisation to become more ‘data-driven’. It is more about changing the culture. For this, the right people are required who can jointly exercise effective influence step-by-step (with pilot cases) to show how and where intelligent use of data can provide additional value. In this way, it is often possible to convince key stakeholders and then to bring ‘data science’ abilities to a higher level through a training programme and/or through the adoption of (extra) data analysts.