How is video consumption developing? What are the main trends? What video channels are most promising? What channels are fading? Mediabrands Marketing Sciences crunched all video consumption data of the Dutch market. Download the complete study [PDF]

Netflix reach grows, Dumpert decreased. Other video sites show stable reach.

NLziet, a Dutch video service set to compete with Netflix, is not in top 10 video site reach. Videoland, bought by RTL to expand OTT presence, also missing.

Yearly comparison monthly reach – NL 16+

Source: GfK DAM - Telecom paper Q4 2015

Video platforms show highest reach in age groups 13-19 and 20-34 years, but 50+ does watch online video.

Netflix 2nd most popular video platform in age group 13-19 years.

Top 10 video sites/apps monthly reach

Source: GfK DAM - October 2015

TV viewing showing downward trend among people <49 years while 50+ years group stabilizes.

2014 was a-typical due to major sports events broadcasted on TV (Olympics and Soccer World Cup).

Time spent watching TV in minutes/day - comparison by target audience

Source: SKO TV viewing, 6 days delay included. 2015 data Jan-Oct

Also less TV watching by people in higher social classes and by shoppers.

Time spent watching TV in minutes/day - comparison by target audience

Source: SKO TV viewing, 6 days delay included. 2015 data Jan-Oct

Downward trend is caused by decline in live TV viewing; delayed TV viewing is only slowly gaining popularity.

Time spent watching TV in minutes/day - comparison by target audience

Source: SKO TV viewing, 2015 data Jan-Oct

If we use a device other than a regular TV for watching AV content, our viewing behavior is completely different.


Devices for watching AV content
Viewing behavior per device

Source: MediaTijd 2014

YouTube is only online video platform comparable to TV channels (in terms of reach), and moving up in rank

Weekly reach AV channels 2014 - NL16+
Weekly reach AV channels 2015 - NL16+

Source: SKO, GfK DAM, October 2015

YouTube weekly reach (site+app) higher among light TV viewers

Weekly reach AV channels 2014 - Light TV viewers NL 16+

Source: SKO, GfK DAM, October 2015

Weekly reach AV channels 2015 - Light TV viewers NL 16+

Source: SKO: Light TV viewer: <=21 hrs/week

  • TV
  • Online devices
  • Mobile devices
  • The graph builder allows you to crunch, slice and dice the raw data we used for our report. Check method for more info.
    Source: GfK Cross Media Link

    Source: GfK Cross Media Link

  • TV
  • Online devices
  • Mobile devices
  • The graph builder allows you to crunch, slice and dice the raw data we used for our report. Check method for more info.
    Source: GfK Cross Media Link

    Source: GfK Cross Media Link

  • TV
  • Online devices
  • Mobile devices
  • The graph builder allows you to crunch, slice and dice the raw data we used for our report. Check method for more info.
    Source: GfK Cross Media Link

  • TV
  • TV youngsters
  • Mobile
  • The graph builder allows you to crunch, slice and dice the raw data we used for our report. Check method for more info.
    Source: GfK Cross Media Link

  • TV
  • TV youngsters
  • Mobile
  • This graph shows the future TV consumption of the 13-16 group of future TV consumers versus the 6+ (all).
    Source: Media Buying Systems (MBS)

  • TV
  • TV youngsters
  • Mobile
  • The graph builder allows you to crunch, slice and dice the raw data we used for our report. Check method for more info.
    Source: GfK Cross Media Link

    About

    This website is a result of a partnership between Mediabrands, GfK and Google.

    Mediabrands Marketing Sciences is part of IPG and the intelligence, data and tech unit of Mediabrands in the Netherlands. Our goal is to measure, predict and optimize the effects of advertising investments of all clients of Mediabrands, including UM, Initiative and Traffic4U.

    GfK is the trusted source of relevant market and consumer information that enables its clients to make smarter decisions. More than 13,000 market research experts combine their passion with GfK’s long-standing data science experience. This allows GfK to deliver vital global insights matched with local market intelligence from more than 100 countries.

    Google’s mission is to organize the world’s information and make it universally accessible and useful. Google provided funding for this study.

    Method

    Executive summary

    Mobile device ownership in The Netherlands has stabilized and almost all Dutch households have a connected or digital TV. This resulted in 2015 to be a turning point for watching linear TV, across all ages. Especially among young people (13-19 years) the declining trend is imminent (-20%). And this young group displays the highest reach for online video content platforms, like YouTube and Netflix. But also people between 35-49 years watch less live TV, and live TV viewing time stabilized for the older 50+ years group. With the current data and research available it is difficult to determine whether the decrease in linear TV viewing has shifted towards online video viewing, as this type of combined data is not available yet.

    Introducion

    In 2014 Mediabrands Marketing Sciences initiated a research study to report a full overview of trends in online video and TV consumption in The Netherlands. The report is based on data gathered by SKO, GfK Cross Media Link, Trends in Digitale Media and other sources. In 2015, we have gathered new data to show the latest trends and developments in video consumption. The study was conducted in collaboration with Google and GfK.

    Data

    In this study we have used a variety of data sources to make a complete overview of TV and online video consumption in The Netherlands:
    - GfK Crossmedia Link by use of DAM and self analysed raw data.
    - SKO (Stichting Kijkonderzoek); SKO is responsible for the reporting and monitoring of TV ratings.
    - GfK Trends in digital media; provides an overview of device ownership in The Netherlands.
    - Media:tijd 2014; provides insight in time spent on various activities such as media.
    - Wave; Mediabrand’s own international social media research

    Data

    GfK Crossmedia Link contains tracking (sensus) data of the online and offline video consumption of around 8.000 Dutch consumers on three types of devices (more info here):
    1. Home TV viewing behaviour measured via Smartphone audiomatching
    2. Software registered home desktop/laptop internet behaviour
    3. Software registered Smartphone/Tablet internet behaviour (iOS and Android)

    More information about this study please contact menno.van.der.steen@mbww.com

    Modeling and number crunching

    The analysts of Mediabrands Marketing Sciences have crunched the enormous ‘GfK Crossmedia Link’ big data set using MySQL and R. Since new devices, operating software and browser software is constantly changing, there were a lot of challenges to manage the analysis of the data. Normally measurement technology follows a few weeks after changes are detected, resulting in a faulty data set during those weeks. We have tried to compensate these faults by leaving this data out of our analysis data set as much as possible. It will show though in the Graph Builder. That is why you will see some strange bumps in the graphs... See the disclaimer for more information.

    Part of this study is a rough prediction of future video consumption between now and 2020. For this we created an econometric model including the influence of seasonality, weather, trends, special events with a big impact on (linear) TV consumption like the Olympics and Soccer Championships. Since models had to be made for each target audience and each channel, we used a stepwise regression for each model. 580 models were estimated with a stepwise regression in both directions. The model selectiono was by AIC. Of course we cannot take new technological developments (like new devices for instance) into account since we don't know there future impact. We used TV and Mobile video consumption data because there were some measurement issues with desktop/laptop viewing data.

    Target Audiences

    In this study we selected a limited amount of 17 different audience groups that are commonly used in marketing and media. Those are based on the following: 16+ M&V, 16+M, 16+V, 20-49M, 20-49V, 20-34M, 20-34V, 35-49M, 35-49V, 50+M, 50+V, 16+ social classes (AB1, B2, C, D) and 20-49 social classes. The reason for the limited amout of segments is to keep the perfomance of this website reasonable.

    Disclaimer

    General remarks

    The study is based on the GfK Crossmedia Link thus for small channels (not viewed by many people), the graphs for those channels might show large fluctuations We worked with capped data which means that viewing times are cut off after 10 minutes for the mobile panel and 12 minutes for the online panel. This leads to underestimation of the online and mobile video consumption for channels like RTL XL, NPO, Netflix…
    The activity of the users is determined differently for the 3 panels:
    1. TV panel: Active between the start month and end month of the person (start month: first month in which active before the 10th of that month, end month: active after 20th)
    2. Mobile panel: Active in a month when active in the first 15 days and active after the 15th. (at least one row in each period)
    3. Online panel: Active in months with a maximum gap of 1 month and the start & end month restrictions corresponding to the TV panel (1).
    All irregularities in the data are shown in the Graph Builder in order to report the data as transparant as possible. It is also very difficult to compensate measurement errors because their impact differ per channel, incident, target audience and time.
    There is a hidden impact due to the challenges of panel management (people leaving and entering the panel, level of participation) that we cannot quantify in this study.

    Data challenges

    The video consumption data of ‘GfK Crossmedia Link’ is measured in different ways. The TV viewing behavior is measured through audio matching technology. At the house of a panel member a box is installed listening to audio from 6.00 AM till 1.00 AM every day. The audio is matched with the broadcast schedules of the pre-defined TV channels. Some of the audio that is detected can't be matched. This data could be a DVD, delayed viewing, etc. We labeled this data 'no match' in the Graph Builder.For this study we removed all data from 6.00-7.00 AM because it resulted in matches that didn't exist: the silence was matched with all kinds of programs in which (sometimes) silence occurs.

    Video viewing online is measured through a software application on the desktop or laptop of the panel member. It registrates all URL's that are visited in the browser that the consumer is using. In August 2014 YouTube switched to a SSL/HTTPS environment which caused that YouTube could not be fully measured. In July 2015 a fix came to hand for Explorer and Chrome which led to more realistic viewing figures.

    Video viewing on mobile devices are measured with an application installed on the device as well (from January till August 2014). IOS8 caused trouble for this methology from the 17th of September 2014 until the 3rd of November 2014. This resulted Gfk to develop a new measurement method. From November 3 GfK started to use a new methodology based on a proxy connected with the router (WIFI) of the household of the panel member. So it doesn't measure mobile video viewing using a mobile network provider but it worked with IOS8. The challenges weren't over though... Lollypop (OS for android) has made it impossible to measure impressions, from the 12th of November 2014. From 11-2014 until 8-2015 missing page impressions from Firefox browser (approximately 15% of panel).

    Predictions

    As everybody already know, but sometimes forgets, is that we cannot predict the future. A model is always a simplification of reality and that also applies to the forecast in this study. The predictions in the Graph Builder can be filtered per channel and per audience type. This way you can play with the future yourself. Bear in mind that these predictions are calculated automatically based on a search algorithm for the best fitting model. We didn't had the time to check all 500+ possible models and their outcomes manually. So you probably will find inconsistencies with your own expectations.

    We only predicted on the dataset of TV and Mobile because of a problem in the measurement of desktop/laptop video consumption in 2014 and 2015.

    Contact

    Please contact us for more information:
    Mediabrands Marketing Sciences
    +31 20 7993100
    Menno.van.der.steen@mbww.com