Google and Facebook are the two largest advertising platforms worldwide and they capture most value of the online advertising market. This duopoly is ripping away value from incumbents as brands are shifting advertising budget from incumbents (telecom operators, broadcasters, …) to make deal directly with Google and Facebook. With such a power of negotiation, is it really a surprise that incumbents are joining forces to propose alternative solutions and bring a credible offer on the market? This has just happened yesterday on the belgian market where the creation of the “Belgian Data Alliance” was announced.
The Belgian Data Alliance is made up of Proximus, DPG Media (VTM, HLN, …), Mediahuis (De Standaard, …), VRT (Dutch-speaking public broadcaster), SBS (VIER), Telenet, Rossel (Le Soir), RTBF (French-speaking public broadcaster) and RTL Belgium.
We’ve seen alliances of smaller sizes between firms in the same sector, international collaboration in the broadcasting world (like the one between SAT1, TF1 and Mediaset) but this type of alliance centred around targeted advertising is a premiere at a national level.
Objectives of the Belgian Data Alliance
The objective of the Belgian Data Alliance (as explained in De Tijd and L’Echo) is enable targeted advertising on digital TV as well as on online platforms (in Belgium an overwhelming part of ad spending still goes to TV. RTBF annual report mentions that digital advertising represents only 3% of the €70,8m advertising revenues for the year 2018). The hope is obvious :
- attract more spending from brands
- earn more with high priced targeted advertising
Technical aspects of the project
The Belgian Data Alliance aims at sharing data they have on users to better understand who they are, what they like and better target them advertising wise. In other words, through data sharing members of the Belgian Data Alliance expect to deliver ads that better suits the profiles and needs of their clients.
The first-party data the different firms involved have on their clients is probably very rich. It includes :
- socio-demographic information (Proximus and Telenet have premium quality data on who you are thanks to the invoicing of their services)
- consumption data
Telenet and Proximus together will probably bring quality data on at least 70% of the Belgian population (Telenet and Proximus cover almost 100% of the Flemish market and VOO has between 30 and 40% of the Walloon market according to IBPT).
The media websites partnering in the alliance are likely to bring consumption data for almost 100% of the Belgian population. The websites involved in the alliance are visited by all Belgians on a regular basis.
All in all, it sounds therefore like a very clever move.
Conclusions
The objectives of the Belgian Data Alliance make perfect business sense. It’s also a question of sovereignty and survival for many firms active in the advertising sector. In Belgium alone Facebook and Google rip off 80% of the advertising market, the size of which is ca. €1bn per year.
Belgian specificities
However I see some hurdles. First of all the potential of the alliance has to be considered taking the peculiarities of the country into account. The French-speaking part of the country is under-represented. VOO, the telecom operator which is the #1 competitor of Proximus in Wallonia is not part of the alliance. This skewed representativeness reminds me of the Media ID venture, a project launched in early 2015 whose aim was to gather users’ data thanks to a unique registration giving access to all Belgian media. Media ID was abandoned 2 years later, one of the reasons being the heterogeneity of the adoption between the northern and southern part of the country.
Richness of users’ profiles
Second, the users’ profiles will never be as rich as on Google or Facebook. Google and Facebook are ecosystems where users are active: they comment, share, like, tag, write and it really makes a difference in terms of profiling. The members of the alliances offer (for the most part) content to consume (read, listen, watch). This is a more passive type of consumption that tells less about the user. But perhaps it’s not really a problem. In the end the age / zip code / gender segmentation is still the most used by digital advertisers and anything richer than that will already be a giant leap forward.
Technical hurdles
Third there are also some big technical hurdles. Sharing data among companies (typically in a DMP) can look like an easy peasy task but it’s not. The more players, the more complicated it will be to set up that project and get it rolling. As long as you inject simple structured data (gender, zip code, age) it may be feasible. But I understand that the project is more ambitious than that. It aims at profiling users based on their consumption (of article, video, audio). This means you’ll have to categorize these different types of media, understand what they are about, analyse which part of it users most stick to. Overcoming such a task requires that you use NLP (Natural Language Processing) to analyse content transcribed into text form (Spech-to-Text) and that you use a common taxonomy. And that is a very different story that is unlikely to be ready for all players alike by the end of 2019.
Credits: image courtesy Shutterstock
Posted in Innovation.
