News & Views

Develop a “Non-Linear” Database

As reported in the October Bulletin, a pilot version of a non-linear database is being developed to potentially accommodate viewing data from non-linear sources that fall outside the core Barb service.

The first experimental output from the pilot non-linear database has been examined by Barb and its research contractors Kantar/TNS and RSMB. The database has proven the concept that data collected from Barb meters in the TV panel can provide detail for purposes beyond Barb’s core service. This offers the potential to identify and report viewing to stand-alone VOD content. Viewing sessions via a TV set, which relate to broadcast television beyond the previous 7 days, can be further analysed and matched to an additional reference bank of material supplied by broadcasters.

The project has so far proved that:

• Barb meter data from all Barb panel homes can be further processed – including viewing via TV sets, accessed from STBs, games consoles, PCs or other forms of playback device;
• instances of viewing to specific content assets held in a reference bank can be identified;
• additional reporting can be structured to these content assets, rather than to a date and time of broadcast.

These are important steps in Barb’s ability to develop reporting for the industry.

Some important questions remain, however – most prominently the identification of surrounding advertising. Some further experimentation is required to understand the possibilities in more detail. Also the issues of processing capacity, volumes of content that can be handled in this way and the interfaces required with content providers. A system would need to be created which has the capability to routinely report based on a changing portfolio of content assets. In addition, far more needs to be understood about the potential volumes of viewing involved and the most logical frequency and means of reporting or summarising such viewing to the industry.

Dialogue with the industry is required to ensure that any development beyond the proof of concept already delivered is able to be considered to be valuable and will meet some industry requirements. If you represent a stakeholder keen to participate in this discussion please e-mail dialogue@barb.co.uk and register your interest.

Pursue a PC/Laptop Test Panel Environment

The field test of the “Virtual Meter”, developed by Kantar/TNS specifically to measure viewing to TV content via PCs and laptops, has commenced in a test panel of 75 homes. The meter is a piece of software, simple to install.

The Virtual Meter test panel is not being undertaken as a quantification of viewing on PCs/laptops but rather as a study to understand the issues involved with this new measurement technology and the feasibility of rolling it out to the Barb panel.

In order for such a technique to be successful it is important to understand the issues in generating and maintaining a sample to participate in measurement – including the use of PCs by multiple household members, the possible numbers of different PCs present, the willingness of respondents to have viewing via PCs monitored and the issues around participation. In addition, the use of PCs for multiple tasks needs greater understanding in order to ensure the greatest confidence in any data returned.

Several well developed QC routines exist in Television Audience Measurement on the Barb panel which have been progressed and enhanced over many years to ensure that the issues impacting on quality of data are addressed. The operation of metering, or tracking of any sort, on PCs/laptops is an underdeveloped field and there is, therefore, much to be examined to understand the data that is returned from such sources. New QC reports are being developed as Barb is aware of many of the issues that need to be considered or confronted. Barb is utilising this project to develop as much understanding as possible about participation, use of devices by respondent households and the nature of the data captured and returned from test panel homes.

The Virtual Meter technique returns two kinds of data associated with access to content – audio signatures and website url data. So far, Barb together with Kantar/TNS has been able to observe that audio signature data can be routinely captured from homes and analysed against TV content references. Webmeter url data is also able, in many cases, to determine or indicate the source of the viewed content, if via a website.

From the additional people information that is returned, a small number of instances of co-viewing can be ascertained but a far larger incidence has been observed of multiple use of the same PC by many different household members at different times. The ability to identify these different types of use gives greater confidence that this form of measurement may be able to provide some valuable insights for the industry.

Early indications suggest the technique could be viable but there remain many issues and questions to probe as the project progresses. The Virtual Meter test panel is envisaged to run until March 2011.

Explore Options to Enhance Server Data with Barb Panel Data

Barb is exploring options to enhance server data with its panel data. To this end, Barb has observer status at meetings of the Broadband Measurement Working Group (BMWG).

Barb will monitor the work of the BMWG to assess whether there are opportunities for Barb, for example options for any development of the non-linear database to be aligned with server data collation of content providers’ usage statistics. It is anticipated that investigations into the possibility for nominated on-demand content, monitored through the Barb system, to be tied to the relevant server data will move beyond a conceptual approach towards experimenting with working applications.

Some dialogue is underway with Barb’s underwriters to test the concepts of aligning server records to test metered usage of laptops and other content services consumed via the TV Set. If viable, this will give greater confidence of the meaning of different definitions of server data and the ability to confirm whether it is feasible to match observed panel member usage (derived from Barb meters) with server records for the activity.

Such a step would offer greater potential for the use of server data to expand on defined types of viewing from the Barb panel survey. This would effectively allow server data to be aligned to the full picture of TV consumption, rather than standing alone unconnected and unrelated to the totality of viewing. This could then provide the potential for server data to operate as a microscope on types of viewing – delivering the detail of served advertising, or of smaller collections of content where niche consumption has not reached critical mass.