Would you like to know more about our plans for the future?
Take a look below at some of these frequently asked questions about Project Dovetail and the TV Player Report.
What is Project Dovetail?
In an era of constant change, BARB continues to develop its services in response to fragmenting behaviour patterns. Since our launch in 1981, there has been proliferation of platforms, channels and catch-up services. In recent years, more people have started to watch television and video content distributed through the internet.
Project Dovetail is at the heart of our development strategy. Its premise is that BARB’s services need to harness the strengths of two complementary data sources.
- BARB’s panel of 5,100 homes provides representative viewing information that delivers programme reach, demographic viewing profiles and measurement of viewers per screen.
- Device-based data from web servers provides granular evidence of how online TV is being watched.
The TV Player Report is the first stage of Project Dovetail.
It is a beta report that is based on the first outputs from BARB working with UK television broadcasters to generate data about all content delivered through the internet. This on-going development delivers a census level dataset that details how different devices are being used to view online TV.
Stage 2 of Project Dovetail will integrate this dataset with the outputs that our customers are used to receiving from BARB. We are in the final stages of tendering the contract for this stage. More news will follow on how we will deliver this.
What is the TV Player Report?
The TV Player Report is the first joint-industry, audited measure of viewing to online TV in the UK.
It reports on the consumption of online TV content in TV players controlled by the broadcasters. Data from all users of these players, not just a panel, go into the report.
We are also reporting on gross live consumption via online.
What does the terminology in the TV Player Report mean?
BARB’s customers are used to working with data that represents the viewing habits of people. The TV Player Report focuses instead on devices.
The core metric for determining the popularity of a programme is Average Programme Streams, which represents the average number of devices that played the programme across its duration. It is analogous to the average audience measure that our customers often use to report programme audiences. Each of these metrics conveys the popularity of a programme, although Average Programme Streams is a measure of device usage rather than a measure of how many people are viewing.
We also use the following terminology in the TV Player Report.
- Total Viewing Time – The total number of minutes that devices have received content (this enables a headline comparison of the volume of delivery through each TV player).
- Android – Viewing to content that is made available through TV player apps that can be downloaded on tablets and smartphones running the Android operating system.
- Apple iOS – Viewing to content that is made available through TV player apps that can be downloaded on Apple tablets, smartphones and other handheld devices.
- Website player – Viewing to content that is made available by broadcasters through their own website (accessible through browsers on multiple devices and operating systems).
Which numbers in the TV Player Report are already covered by BARB's established panel data?
Currently BARB only reports online TV viewing when PCs, tablets or smartphones are being cast or projected onto a panel member’s TV screen. As such some of the viewing reported in the TV Player Report is currently included in the viewing estimates provided from the BARB panel.
Are the data generated only from BARB panel homes?
The viewing levels are generated from across the whole population, not just from our nationally representative panel of homes. This census-level approach provides great certainty in overall viewing levels. The viewing data in this report do not represent the number of people that are watching. Instead, the figures are based on device usage.
What is Average Programme Streams?
Average programme streams is an industry-agreed metric that indicates the popularity of an online TV programme. It’s the total viewing time of a programme on a TV Player by all devices in the reported period, divided by the full length of the programme.
It is analogous to the average audience measure that our customers often use to report programme audiences. Each of these metrics conveys the popularity of a programme, although average programme streams is a measure of device usage rather than a measure of how many people are viewing.
The average programme streams metric has been ratified by JICWEBS, the industry body responsible for the development of standards in measuring online media performance.
Why is the TV Player Report being published in beta format?
There are three areas of continuing development that will turn the TV Player Report into an increasingly comprehensive picture of online TV viewing.
BARB, with the research agency Kantar, is working with broadcasters to implement the analytics tag into more and more TV player platforms. This has to be done platform by platform. The report will include each platform as soon as its tagging implementation has been independently audited.
The second area of ongoing development is identifying which programmes are being watched live. The analytics tag can determine when someone has chosen to use a TV player to watch a channel live, although it can’t access content IDs that identify which particular programme was on at the time. A process is being developed that matches the timestamp of the viewing session to the programme logs for the channel being viewed.
Another important development is the attribution of on-demand viewing back to the originating broadcaster, when this viewing takes place through a TV player that aggregates content from different broadcasters (e.g. Sky Go). We can already identify the channels streamed live through Sky Go and are working on a solution that will allocate the time spent viewing on-demand content back to the originating broadcaster.
Why are some broadcasters' platforms missing from the report?
Each broadcaster has to schedule the development work in alongside other software development priorities. Many broadcasters have started implementing the software code which means that the report will provide an increasingly comprehensive picture of online TV viewing as time goes on. A number of implementations are currently being audited, which is the final stage of preparation prior to reporting.
BARB will publish data from TV players that have passed an independent audit by ABC (Audit Bureau of Circulations). This gives us added reassurance that the analytics are performing as expected and the numbers make sense.
Does the report cover live streaming and on-demand viewing?
Yes, it is designed to report online minutes for both types of viewing, although the following needs to be considered when reading the report.
Firstly, the specific content watched is already identified for on-demand viewing through a system of content IDs. We will be able to report which programmes were viewed live by integrating BARB’s comprehensive, as-run programme data logs. Consequently, we will report average programme streams for both live and on-demand viewing in due course.
Secondly, some TV players are still developing the ability to deliver data on live streaming. The delivery of on-demand content and live streaming are often separate technical processes within a TV player. Each of these requires distinct coding, implementation and auditing.
Are you reporting on consumption of advertising?
We are developing this and will report it in due course. Serving and identifying ads on the internet is more complex than programme content, with multiple suppliers and technologies involved.
Is viewing to downloads included?
Yes, offline viewing of downloaded content can be captured for broadcasters that have enabled this feature in their TV players. When this feature is enabled, the data for offline viewing of downloaded content are retrieved the next time a device goes online.
What period does the TV Player Report cover?
The report is published on a Wednesday and covers the calendar week that ended ten days previously. It also includes rolling cumulative information for the last four weeks. It is published on the home page of the BARB website (www.barb.co.uk).
What proportion of all online viewing does the report account for?
There is no accepted currency that covers all forms of online TV and video. That said, broadcasters’ own data suggest that as much as 80% of online TV viewing is covered by Android apps, Apple iOS apps and Website players.
What are Android, Apple iOS and Website Players?
These are internet platforms on which broadcasters have developed TV players. The Android figures reflect usage of TV Player apps developed for this operating system that is used on a range of tablets and smartphones, while the Apple iOS data cover viewing that takes place on TV player apps developed for use on iPhones, iPads and iPod Touches.
Website players are found on broadcaster websites which can be accessed through a variety of browsers on multiple devices and operating systems.
Will the report encompass other operating systems?
Yes, we intend to add other online distribution platforms in due course.
Does the TV Player Report include viewing through internet-only content platforms?
The methodology is designed to be inclusive for content distributed through online TV players. BARB has no intention to restrict inclusion in the report to its current channel subscribers.
How does the implementation process work?
Kantar provides the software code for the analytics tags to broadcasters in an SDK (Software Development Kit). The broadcaster’s app development teams implement this code within each of their TV player apps.
Having done this, the broadcaster works with Kantar on a series of acceptance tests to ensure that data are being generated in line with BARB’s expectations.
How does the auditing process work?
Once the broadcaster and Kantar are satisfied that the software code (SDK) has been implemented properly, the player is ready for audit.
During the designated audit period, ABC watches random on-demand and live content on appropriate devices and checks that what’s reported by the broadcaster aligns with what it expects. It also seeks reassurance that there is no robotic traffic going into the reports, and that the overall numbers claimed for the period can be substantiated by the data supplied. All of this aims to prove that the figures reported make sense and reflect genuine human activity.
Once ABC is satisfied with all its tests, the audit is passed and the player goes live for reporting. Once a player is live, random audits are scheduled during the year to ensure that we can still have confidence in the system. A new audit will be commissioned if a broadcaster makes a major update to a player, such that it’s effectively a new piece of technology.
Has BARB had any issues collecting data in line with its industry-agreed standards?
There are occasions when viewing data and the allocation of programme names have been inaccurate. BARB works with broadcasters and Kantar Media to resolve these as swiftly as possible and may take the decision to remove data from reports. The table below highlights known issues with the data; we advise you to take these into account when looking at the TV Player Report.
|Date From||Date To||Resolved?||Players Affected||Effect of issue on reporting|
|Feb-18||Mar-18||No||C4 Android app||Live streaming play time on the C4 Android app was over-stated.|
|Feb-18||Feb-18||Yes||BBC website player||During the Winter Olympics, live streaming was only measured for the event leading to overall under-statement of live streaming play time.|
|Dec-17||No||Sky Go Android app||Download measurement has been switched off in the app, slightly reducing overall measured play time.|
|Nov-17||Jan-18||Yes||UKTV website player||Overall UKTV Play website play time and Average Programme Streams have been over-stated for this period. A new version of the player has resolved the issue.|
|Sep-17||No||My5 iOS app||A few My5 programmes have wrong Average Programme Stream numbers due to their durations being passed incorrectly. This may affect My5 programme rankings. Total play time is not affected.|
|Dec-17||No||Sky Go Android app||Overall play time since December 2017 will be reduced slightly because download measurement has been switched off in the app.
|Aug-17||No||My5 website player||Overall My5 website play time and Average Programme Streams have been under-stated for this period. The My5 website player has been withdrawn from public reporting until this issue is resolved.
|Jul-17||No||Sky Go Android app||Around 1% of Sky Go play time is showing as "Not attributable" to an originating broadcaster.
|Apr-17||Jun-17||Yes||Sky Go Android app||Overall Sky Go Android play time and Average Programme Streams for this period were under-stated.
|Mar-17||Sep-17||Yes||BBC iPlayer iOS app||Play time and average programme streams were overstated during this period on BBC iOS. The issue affected viewing sessions of more than one hour, principally on live streaming.
|Mar-17||No||All STV players||Rankings tables for STV content may show "Programme Name Not Available".
|Feb-17||Sep-17||Yes||STV on-demand players||Overall playtime for STV on-demand players has been over-stated due to an aggregation error. Rankings and Average Programme Streams are not affected.
|Dec-16||Dec-16||Yes||All ITV players||Rankings tables for ITV content over this period may show strange-looking programme titles.
|Dec-16||Jan-17||Yes||All STV players||The same programme title may appear more than once in rankings over this period due to multiple versions being available.
|Nov-16||Feb-17||Yes||All Sky Go players||The team names for live Premier League matches over this period may be stated incorrectly.
|Oct-16||May-17||Yes||All ITV players||The same programme title may appear more than once in rankings over this period due to multiple versions being available.
|Oct-16||Nov-16||Yes||ITV iOS app||Live streaming play time on ITV Hub for iOS was over-stated.
|Oct-15||Dec-15||Yes||All BBC players||The same programme title may appear more than once in rankings over this period due to multiple versions being available.
|Jun-15||Jan-16||Yes||BBC iOS app||Average Programme Streams for BBC iOS on-demand programmes over this period had to be calculated manually.
Why is the total viewing time for TV players and broadcaster groups not exactly the same as the combined viewing time across different devices?
There will typically be a small discrepancy (around 0.2%) between a player’s aggregate overall play time and the sum of its play time across devices for the same period. These numbers come from the same data, which undergo different compilation processes.