What has changed in how BARB reports television viewing? People are increasingly watching television programmes on devices other than the TV set, be it on PCs, tablets or smartphones; the UK television and advertising industry needs a trusted, independent source of these multiple-screen viewing figures. BARB established Project Dovetail to meet this need. For more information on Project Dovetail, including our method for producing multiple-screen programme viewing figures, please visit the Project Dovetail overview page. How were multiple-screen programme viewing figures released? How do I access them? Multiple-screen programme viewing figures were delivered in three stages: DataAvailability on BARB website Availability for BARB customersPublication date 1Collecting and reporting device-based census dataBVOD services report includes viewing time via all BARB-reported BVOD services on PCs, tablets and smartphones.As per website data.2015 2Programme average audiences - For all individuals across four screens: TV sets, tablets, PCs and smartphones - Published 8 days after transmission Weekly top 15 programmes by broadcaster group/channel on four screens via new dashboard in viewing data section. Customer log-in to access weekly top 50 programmes by broadcaster group/channel on four screens via dashboard. Programme data and three-screen demographic analysis (not smartphones) available via TechEdge; other bureaux can provide equivalent analyses.25 Sept 2018 with back-dated figures for programmes broadcast from 27 August 2018 3Multiple-screen programme reach and time spent viewing - For viewing on three screens: TV sets, tablets and PCsUnduplicated multiple-screen programme reach and time spent viewing on three screens by broadcaster group/channel. This total three-screen viewing is reported alongside consolidated seven-day TV set viewing.In addition to the multiple-screen programme reach and time spent viewing data published on the BARB website, respondent-level data are available to bureaux, which allow more extensive analysis of unduplicated reach across multiple screens for a range of demographic groups.21 January 2020 How does BARB produce multiple-screen programme viewing figures? BARB has two sources of data. BARB collects device-based data for online TV viewing whenever anyone in the UK watches television via a tagged broadcaster video-on-demand (BVOD) service. Viewing data generated from this census-level dataset are based on device usage and do not represent the number of people that are watching. We collect observational data continuously from our nationally-representative panel of 5,300 UK homes. Over 60% of these homes have had software meters installed that track our panel members’ viewing on tablets and PCs. We know how individuals view across different screens because we measure viewing on all devices in the same panel. To generate multiple-screen programme viewing figures, we need to understand two aspects of online TV viewing from the panel: The extent to which people watch BVOD services on their own, or as part of a group; this is known as viewers-per-view. What type of people watch BVOD services on different devices. Simply put, we are then able to turn device-based data into people-based viewing figures by aligning programme audience levels to the census data, while aligning programme audience profiles to the panel data. This results in multiple-screen viewing figures. Read this BARB Explained to find out more. When will BVOD campaign performance be available? The fourth and final stage of Project Dovetail is BVOD campaign performance (multiple-screen commercial viewing figures). In June 2020, we launched a beta BVOD campaign planner to help agencies and advertisers plan advertising campaigns across BARB-reported commercial broadcasters’ linear channels and VOD platforms. This is the first joint-industry tool designed specifically to plan total TV campaign performance. All data integration projects have to contend with the statistical effects of combining two different data sets. Our development work indicates that the effect of using the Dovetail Fusion methodology for reporting total TV campaign performance carries the risk of introducing a level of reach inflation. Due to this, we are working with our stakeholders to see if there’s another way to best meet the industry need for post-campaign evaluation across multiple screens. We expect to announce more detail on this in early 2021. What new type of multiple-screen programme viewing reporting did BARB launch in January 2020? On January 21st 2020, BARB began to report unduplicated programme reach and time spent viewing figures on a daily basis across three screens (TV sets, tablets and PCs). What has changed in how BARB reports reach and time spent viewing programmes on its website? The website features a new weekly viewing summary report, which provides channel and PC/tablet reach, share and time spent viewing data based on BARB’s new definition of total TV, total three-screen viewing, that takes place within a rolling four-week period. Weekly data based on the previous definition of total TV, C7 TV set viewing (consolidated seven-day TV set viewing) are reported alongside for continuity. What new analyses can be done with the multiple-screen reach and time spent viewing programme data? Reach analyses for computers/tablets, and incremental reach analyses for these devices over TV sets. Please refer to the PVX – recommended analyses document for a comprehensive list of the types of analysis that are supported by the new data. Why are there limitations to the recommended analyses? A sophisticated model is used to produce the PC and tablet reach figures, which differs from the way in which the existing TV set reach figures are produced. The modelling process introduces compromises which mean that in some cases, users cannot conduct the same type of analysis on the PC and tablet data as they do on TV set data. Why is 28-day reach a recommended analysis, when daily and 7-day reach are not? When building the model that produces the PC and tablet reach figures, we had to prioritise one type of reach estimate. The modelled dataset used is designed to achieve the most accurate estimates of reach over a 28-day period. Can genre analysis be conducted? Genre analysis can be conducted on a time spent viewing basis, but is not recommended for reach analysis that includes PCs and tablets. How is viewing of third-party broadcasters’ content on tablets and PCs via other BVOD services reported e.g. ITV programmes viewed on Sky Go? In the new weekly viewing summary report, viewing via tablets/PCs has been attributed to the originating broadcaster, i.e. ITV programmes viewed on Sky Go have been attributed to ITV. Which BVOD services are included in the four-screen viewing dashboard? The four-screen viewing dashboard includes multiple-screen viewing figures for tagged BVOD services only. At present, these are: All4, BBC iPlayer, ITV Hub, My5, S4C Clic, Sky Go, STV Player and UKTV Play. What are the known issues with the multiple-screen programme viewing data in the four-screen viewing dashboard? PC, tablet and smartphone viewing figures are only counted where the broadcaster has supplied consistent content identifiers, which link online viewing to the linear broadcast. The following known issues also affect the four-screen viewing data in this dashboard: Online data are not available from the following BVOD services: STV web and iOS live, S4C Clic Android on-demand / live. This reduces the tablet /smartphone viewing reported via the STV Player and S4C Clic services. Viewing data on PCs, tablets and smartphones are currently not available for the following broadcasters/channels: Asia TV; CBS AMC; E!; Eurosport; MSM Asia; Star India; VIMN (MTV, Comedy Central, Nickelodeon). We are working to resolve these issues and to make these data available in the dashboard as soon as possible. I’m a BARB customer; what new data files are being produced and how do I access them? BARB is producing two new primary file outputs: A complete daily audience file of all online TV programmes, broken down by the most widely-used demographic groups. This file is called web database 2 (web db2). Programme average audiences can be calculated from this file. A daily respondent-level data file, with top programmes and broadcaster/player audiences calibrated. Unduplicated multiple-screen reach and time spent viewing can be calculated from this file, which is called the cross-platform panel viewing file (pvx). These files are available for BARB customers and all BARB-registered data-processing bureaux. Please ask your bureau for more information on which data it is making available for users. Why is smartphone viewing only available for all individuals in programme average audiences? Converting device ratings to people ratings requires measurement of how our panel members view BVOD services on different devices. Router meters will enable us to measure our panel members’ viewing on smartphones. We have commissioned Kantar to install its router meter technology into the BARB panel of homes. Kantar began to install its router meters into new and existing BARB panel homes in October 2019, and all eligible homes are expected to have had a router meter installed by the end of 2020. The initial publication date for data from the meters will be confirmed in due course. Meanwhile, we can include programme average audience figures for smartphones alongside other screens for all individuals. We do this by assuming that each smartphone viewing session is viewed by an individual of unknown demography, i.e. one viewer-per-view. Smartphone viewing will not be included in multiple-screen reach and time spent viewing until we have a panel measurement of viewing on smartphones. How has the BARB universe changed with the launch of multiple-screen viewing figures? BARB’s universe has been private households in the UK with at least one TV set. We have extended the universe to include private households with a broadband connection and no TV set; we measure viewing on tablets and PCs in these homes. Broadband-only homes add approximately 3% to the universe. For continuity, users can still analyse viewing within the TV-household universe. For reference, BARB’s universe does not include student halls of residence, military bases, offices and prisons. What's the definition of total TV? Has this changed with the launch of multiple-screen viewing figures? The previous definition of total TV is now called C7 TV set viewing (consolidated seven-day TV set viewing); for continuity, channel reach and time spent viewing are published to this definition on BARB’s website, alongside a new definition of total TV. The new definition is called total three-screen viewing. This encompasses the new forms of viewing that BARB reports. Timeshift viewing on TV sets (either via a recording device or via a BVOD service) up to 28 days after broadcast. Viewing to programmes that haven’t featured in a linear schedule (i.e. BVOD viewing pre-broadcast, or BVOD-only programmes). Viewing on tablets and PCs. Viewing in private homes that don’t have a TV set but do have a broadband connection. Total three-screen viewing was introduced at the same time as the launch of multiple-screen reach and time spent viewing. Users have the flexibility to create their own definitions through analytics systems. Smartphones aren’t yet included in the new definition as we don’t have a panel measurement of how people watch on these devices. For reference, census data show that smartphones add approximately 0.25% to total viewing time. What is the impact of the new definition of total TV on reach? Total three-screen viewing, the new definition of total TV, includes homes that don’t have a TV set but do have a broadband connection. People in these homes are less likely to be TV viewers as they don’t have a TV set; we measure their viewing on tablets and PCs. There is an increase in the actual number of viewers when we include broadband-only homes in the universe. But this increase in the number of viewers is not as big as the increase in number of people in the universe; this is why the proportion of people viewing is lower than in the previous definition of total TV. What is the difference between As Broadcast and As Viewed? Previously, BARB reported programmes as being watched at the time they were broadcast, even if the viewing was timeshifted. This is called the As Broadcast method of reporting. The As Viewed method of reporting allocates viewing to the time it was actually watched. Multiple-screen reach and time spent viewing figures are based on total three-screen viewing, the new definition of total TV, and are reported on the BARB website using the As Viewed method. This is because they include timeshift viewing up to 28 days after broadcast, and we don’t want to slow down our reporting of reach and time spent viewing. We continue to use the As Broadcast method for reporting programme audiences in the dashboard on our website. We also use the As Broadcast method in the reporting of C7 TV set viewing (the previous definition of total TV which is available on the website for continuity purposes). Data-processing bureaux may allow analysis of viewing based on either definition. What is the new BARB gold standard? Gold standard refers to the calculation methodologies agreed by BARB. All the viewing figures we report are a result of implementing these gold standard methodologies. There is no change to the industry-agreed default position for the trading currency. This is consolidated seven-day commercial impacts on TV sets. We have changed the definition of total TV that’s used to report multiple-screen reach and time spent viewing. The new definition, total three-screen viewing, encompasses the new forms of viewing that BARB reports: tablets and PCs, non-linear programme box sets, 8-28 day catch-up and viewing in broadband-only homes. Why don't some BARB-reported channels have multiple-screen viewing figures? We can only produce multiple-screen viewing figures for BVOD services that have implemented software code to our specification; this is known as tagging. This software code generates the census data required from tablets, PCs and smartphones. We cannot collect census data from non-tagged BVOD services. Where channels are available to view through more than one BVOD service, we are only able to report multiple-screen viewing figures for the tagged service. For example, a channel with its own non-tagged BVOD service and which is also available through the Sky Go and Virgin services, will only have its multiple-screen viewing figures via Sky Go reported, as this is the only tagged service. We currently produce multiple-screen viewing figures for the following BVOD services: All4, BBC iPlayer, ITV Hub, My5, S4C Clic, Sky Go, STV Player and UKTV Play. I operate a BARB-reported channel with a BVOD service, how do I track multiple-screen viewing data? We produce multiple-screen viewing figures for BVOD services that have implemented software code to our specification. If you are interested in tagging your BVOD service in this way, please contact us. How much does BVOD viewing on tablets, PCs and smartphones add to TV set viewing? Collectively, time spent viewing BVOD services on tablets, PCs and smartphones adds approximately 1.3% to TV set viewing. This is an average across all programmes, some of which have a much greater proportion of their audiences watching on devices other than the TV set. What is non-linear viewing and where is it included? Broadcasters are increasingly making programmes available to watch on their BVOD services before (or without) linear broadcast. BARB can report audiences to non-linear programmes with the co-operation of the broadcaster. In these cases, non-linear viewing is included in total three-screen viewing, our new definition of total TV. See this BARB Explained for more information. Where is BVOD viewing that takes place via the TV set included? When a panel home watches a programme that was broadcast within the last 28 days on a TV set, this viewing is included in the programme average audience figures, even if they have watched through a BVOD service. This is because we use audio matching to identify which programmes are watched on a TV set. We can’t yet allocate timeshifted viewing to a BVOD service or to a recorded playback, except in Sky homes. See this BARB Explained for more information. We don’t yet report viewing to BVOD commercials on the TV set. How does BARB define viewability on PCs, tablets and smartphones? BARB’s principal convention is that a television programme or commercial has been seen when it is watched at normal speed, with at least one panel member registered as being in the room. Importantly, we deliver duration-based metrics. These conventions guide us, as we plan to report viewing of online TV delivered on TV sets. This BARB Explained explains our approach to the issue of whether programmes and commercials are visible on the screen. Is BARB's online measurement vulnerable to click fraud? We commission independent auditing to reinforce the trustworthiness of our published viewing data. The audit process verifies that data we receive has removed three types of activity: a) viewing from IP addresses that are located on the business premises of the broadcasters whose programmes are being measured, b) non-human traffic, and c) traffic from outside the UK. The audit process is conducted by ABC (Audit Bureau of Circulations).