egta insight online audio hybrid audience measurement January 2016

www.egta.com / / Table of contents

5 Chapter 1: The online audio industry: publisher categories and characteristics

8 Chapter 2: Measurement challenges: quantifying audiences

12 Chapter 3: Overview of audio measurement methodologies

16 Chapter 4: about egta Strategies for measuring audio audiences

22 Chapter 5: egta is the Brussels-based Hybrid measurement: merging terrestrial and online audio measurement and metrics trade association of more than 120 television and radio 23 Chapter 6: advertising sales houses. A look ahead egta’s members are spread across 39 countries, mainly 24 Market spotlight in Europe. Together, egta’s TV members represent over 80% France of the European television Germany advertising market, whilst egta US radio members collect 60% of radio advertising revenues in 38 Glossary countries where they are active. 39 References As sales houses of both public 40 Acknowledgements and private broadcasters, egta members commercialise the advertising space around audiovisual content available on platforms such as traditional radio sets, tablets, smartphones, PCs, Smart TVs and other Internet-connected devices. www.egta.com Page 2 Page 3 parent station. For public service broadcasters, executive summary Chapter 1 which often feature talk and sport programming on some of their stations, online audio extends The online audio the accessibility of digital-only services by adding The online audio industry comprises a wide va- industry: publisher a complementary layer to digital terrestrial broad- riety of publishers, characterised by different cast (DAB, DAB+). distribution and monetisation strategies. Under- categories and The business model behind online radio is similar standing and analysing the audiences of online characteristics to terrestrial radio; it is funded by either advertising audio is currently achieved by a combination of or some form of subscription, licence fee/taxation, research methods, including existing RAM (Ra- The Internet has played host to an ever increasing or a combination of the three. Broadcasters may dio Audience Measurement) methodologies, variety of audio services, from the first online radio choose to sell their online radio advertising inven- IAM (Internet Audience Measurement) and other stations of the early 1990s, through near ubiqui- tory as a package with their terrestrial inventory, techniques such as discrete surveys. tous simulcast streaming of terrestrial stations, to today’s rich landscape of linear and on-demand or they may sell it separately on an impression-by- Whilst it is difficult to accurately assess the com- services. Online audio has grown to complement – impression basis as part of their wider online of- parative scale of online audio listenership across and in some cases incrementally replace – exist- fer. In Europe, simulcast radio streams are usually markets, audiences are clearly growing. Radio ing technologies, such as physical recordings and identical to the broadcast stream, with the origi- broadcasters are reaching ever more listeners terrestrial radio. Rapid increases in fast broadband nal advertising maintained. Online listeners are in online, including on mobile devices, and on-de- penetration, the emergence of legal music stream- most cases accounted for using the radio audience mand music services are increasingly popular. ing services as a replacement for illegal file-sharing measurement (RAM) methodology used to pro- Reports published in markets such as France, platforms and advances in mobile Internet have all duce the ratings on which advertising is traded. In Germany, the Netherlands, Switzerland, the UK contributed to the success of online audio, which some cases, for example France, Spain and the UK, and the US offer insights into these trends. has lead to disruption in both advertising funded it is possible to identify what proportion of a sta- and consumer retail markets. tion’s listeners are tuning in, for example, on FM, Sample-based and census-level measurement DAB, via the Internet, etc., whilst many markets do techniques are being deployed to measure Broadly speaking, publishers active in the online not have this differentiation. In the US, however, online audio listening, and each of these has audio industry can be classified into four groups: it is commonplace to substitute the advertising certain capabilities and limitations. Sample- online radio, which includes live simulcasts of in the online stream with different spots, a phe- based measurement can provide unduplicated terrestrial stations, brand extensions and linear nomenon that largely arose through the historical estimates of reach and frequency, with demo- pure-play Internet broadcasters (that do not offer contractual difficulties of using advertising copy for graphic and geographic information appended, playlist personalisation); online radio aggregators; both broadcast and online (see page 34 for further although such methods decrease in accuracy for on-demand music/audio services, including video details). small and niche publishers. platforms and podcasting services; and person- alised radio services. / / Online radio aggregators Census-level measurement provides accurate data on total usage in terms of contacts and / / Online radio Online radio aggregators blur the lines between the user interactions, but additional techniques are concepts of broadcasters and platforms. Broadly needed in order to provide demographic or other Simulcasting terrestrial radio online is practically speaking, aggregators fall into three types: servic- forms of information about listeners. ubiquitous, and for the most part it allows stations es developed by a single broadcaster for their own to be listened to well beyond their broadcast areas, channels (e.g. iHeartRadio, which also hosts some Hybrid methodologies are currently being de- including by audiences abroad and on new con- other broadcasters’ channels); services that pres- veloped that combine sample and census mea- nected devices. Many broadcasters offer a range of ent several broadcasters’ stations from a single surement in order to deliver the next generation brand extensions, which tend to be curated playl- market together (e.g. the UK Radioplayer); and in- of audio audience measurement. ists catering to more niche musical tastes than the dependent services that aggregate stations and –

Page 4 Page 5 in some cases – on-demand content from a wider music libraries, with little differentiation in terms geographical area (e.g. TuneIn Radio). of pricing plans. figure 1: The more sophisticated online radio aggregators Major players include Spotify, Apple Music, Deezer, Characteristics of online audio publishers offer both live radio and on-demand listening of Google Play Music, Rhapsody, Microsoft’s Groove archived content. Personalisation and recom- (formerly Zune Music) and – the most recent re- mendation features are also to be found on some launch – TIDAL. The market has seen consolida- On-demand Online radio Online radio Personalised aggregators. Typically, versions exist for desktop, tion over the past few years, with Rdio (Pandora), music/audio stations aggregators radio services smartphone and tablet operating systems, and in Napster (Rhapsody) and Songza (Google) all being services some cases other digital devices, such as games acquired by bigger companies and others such as Spotify TuneIn Radio Deezer consoles, wireless Hi-Fi systems and over-the-top Grooveshark closing completely. radionomy Google Play Music Pandora Radio iHeart Radio (OTT) television appliances. FM/AM/DAB+ Apple Music AUPEO! The business model for on-demand music ser- UK Radioplayer broadcasters TIDAL Last.fm Examples of publishers and their brand Radio.de For broadcasters, revenue opportunities are simi- vices typically combines free (ad-supported) and extensions Rhapsody Jango maRadio.be Guvera lar to simulcast or online-only radio. However, the premium (subscription) tiers. These services are Radiocircus Groove customisable display advertising formats of ag- not included in national radio ratings; on the one Karnaval.com Acast gregators such as the Radioplayer allow synchro- hand, radio broadcasters may be wary of including nised visual and interactive advertising to be used competitor services in the RAM, and on the other Terrestrial radio simulcast     as a complement to audio within a standardised hand on-demand audio companies operate a very framework. Linear content accessed through ag- different business model to free-to-air radio. Radio brand extensions     gregators may be included within existing RAM / / Personalised radio services methodologies, although independent third-party   services and on-demand services present a great- Curated content   Personalised radio services typically allow us- (in some cases) (in some cases) er challenge for measurement purposes. ers to generate playlists that cater to their indi- Radio programme archive     Market-level aggregators offer a clear advantage vidual tastes, taking as a starting point either a (in some cases) to the broadcasters present on them: they can in- genre, artist or song title. Users may influence crease their reach and therefore the scale and rev- subsequent song selection by giving a thumbs up Advertising funded     enue opportunity of the online audio advertising or thumbs down to the track currently playing. It offer, and they allow the radio industry to present should be noted that the music libraries of person- Subscription funded     (in some cases) a united voice to other industries, such as automo- alised radio services are much smaller than those of on-demand audio providers, and they may be tive manufacturers. Whilst broadcasters may take Log-in required     the view that they do not want to appear on third- considered a more direct competitor to traditional (very rarely) (in some cases) party aggregators such as TuneIn, these services broadcast radio. Song skipping     do offer an additional opportunity for listeners to The most successful actor in personalised online find and access their online streams. radio is Pandora Radio, founded in 2000 and now Song likes or dislikes     / / On-demand audio services the largest audio streaming platform by registered (very rarely) users. Operating in the US, Australia and New Zea- Individual track selection Services that offer on-demand, legal access to land, Pandora offers both advertising-supported     audio content, predominantly music, have grown and paid (ad-free) tiers, with the latter set at Typical size of music >10k songs rapidly over the past few years, having a major im- roughly half the price of on-demand services such n/a ~ 30m songs ~ 1m songs catalogue monthly pact on the way people access and consume mu- as Spotify and Deezer. Roughly 80% Pandora’s rev- sic. Competition in this sector is fierce, with several enues are derived from advertising1, and the com- companies offering relatively similar services and pany has been successful in growing its mobile

Page 6 Page 7 and local advertising income in particular. Whereas Chapter 2 / / Measurement challenges: Spotify makes about 90% of its revenues from sub- capturing audiences across scribers, a much smaller percentage of Pandora’s user base is signed up for the paid tier. Measurement all platforms and devices and developing convergent In Europe, a Pandora-like start-up called AUPEO! challenges: quantifying currencies Personal Radio launched in 2008 and by 2014 was audiences available in over 40 countries worldwide. Offering a The primary challenge in measuring online audio Quantifying the amount of online audio listening similar listener experience to its American counter- is determining – and implementing – a technical that is taking place, and the effect that may have part, AUPEO! also operates a two-tier model. solution that covers the medium in its entirety. on other media consumption patterns, is problem- Terrestrial radio simulcasts, online radios, aggre- Personalised radio services do not appear in na- atic for three reasons. gators, on-demand services, catch-up platforms, tional RAM, although some do appear in Triton podcasts, music video players and audiobooks, to Digital’s Webcast Metrics (see page 31). Firstly, there are myriad different definitions pos- sible, and these may include all or only part of the cover just some of the array of publishers in online / / Podcasting and on-demand types of service outlined above. Further to that, a audio, employ different software and technolo- / / Towards a new currency for gies to deliver content, some of which is live and spoken word audio significant volume of online video viewing, such as audio music videos on YouTube, could also be classified some time-shifted. Individual audio channels, for Podcasts have been a popular, if relatively niche, as online music streaming and a potential replace- example terrestrial radio online simulcasts, may To date, three solutions have been found to part of the online audio landscape for a number ment for platforms such as broadcast radio or be accessible via several platforms including the address the issue of currencies. The first, and of years. There are currently over 300,000 pod- other types of music consumption. radio station’s own audio player and various ag- simplest, is to limit the currency to terrestrial casts on the iTunes directory alone. While barriers gregators. Without a direct first-party measure- radio and its simulcasts only, using sample- Secondly, continuous and standardised market- to entry are low, barriers to scale and profitability ment mechanism, such as user registration, these based measurement that includes the whole level measurement is in its infancy and the data are high. Therefore, most podcasts remain small as publishers may be unable to monetise a significant audience, regardless of the device used to available is typically very fragmented. To estimate they do not have the advantage of network effect, part of their inventory. consume the content. Under this approach, the scale of total online audio listening depends on editorial and talent development or monetisation advertising on online-only channels is sold on stitching together a range of different reports and Furthermore, there exists a wide variety of busi- possibilities. an impressions-based basis. studies, with different methodologies used from ness models and geographical restrictions that The biggest challenge to podcast profitability is country to country. dictate how content is consumed and monetised. The second, as seen in the US (see page 30) measurement. Podcast programs can be down- Merely finding agreement between these different is to develop parallel and broadly equivalent And thirdly, subscription or registration-based au- loaded or streamed, and there is no standardised companies for a common measurement meth- measurement and currencies for terrestrial dio streaming companies do not publish traditional measurement system across publishers. As a re- odology to cover all audio can be problematic, let radio and online audio, with the latter includ- consumption statistics such as Reach, TSL or AQH sult, podcasting lags streaming music in precise alone designing the technical infrastructure to de- ing a wider range of publishers. consumption metrics and third-party verification. and typically only share information about sub- liver reliable audience data. 80% of podcast consumption takes place on iTunes, scription base size or app downloads. And while The third, as exemplified by the convergent The massive difference in scale of terrestrial radio where publishers have no access to data, and there such services can at least track various statistics currency project in Germany (see page 26), is audiences and those of most online audio pub- is no universal agreement among buyers over about their users for themselves and their clients, to develop a hybrid measurement methodol- lishers also presents a challenge for building an what kind of metrics they require for podcasting. they do not necessarily have any commercial inter- ogy to cover potentially all forms of online est in participating in market-level research. appropriate currency to holistically serve all audio audio. At this early stage, such a converged advertising. FM radio stations, for example, are currency cannot achieve the granularity of ter- measured to a high degree of granularity (typi- restrial radio, due to the small individual audi- cally to quarter-hour day parts, or at the minute ences of online audio publishers. level in the case of some PPM markets), albeit with relatively infrequent reporting (typically by month, quarter or half year). The large and stable audienc-

Page 8 Page 9 Share of radio listening in the Netherlands by figure 2: device-type and target group3

online audio data from four markets Total

13-34 yrs Americans’ share of time spent 15% 1.7% 1.5% 1 listening to audio sources 34-49 yrs 19% AM/FM 21% 10% 50 yrs and over 5.2% AM/FM Radio Cable 7.7% Owned Music (CDs, digital music files, etc.) Internet Internet Radio/Music 11.6% (Pandora, Spotify, etc.) DAB+

52.1% SiriusXM Don’t know

TV Music Channels

20.3% Weekly reach of digital streaming Podcasts services in Germany4 8.8% Other Share of streaming 7.2% audio in the UK5 Monthly online radio listening in the US2 53%

47% 45% 2.9% 2.8% 39% 34% 43%

27% 27%

21% 20% 21% 57% 17% 16% 15% 12% Share of live radio listening by 6 7% device-type in the UK 5%

Radio Brands 2000 ‘01 ‘02 ‘03 ‘04 ‘05 ‘06 ‘07 ‘08 ‘09 ‘10 ‘11 ‘12 ‘13 ‘14 ‘15

Online Radio = Listening to AM/FM radio stations online and/or listening to streamed audio content avail- Music Streaming Services able only on the Internet

Sources:

1: Edison Research, 2014. 2: Edison Research/Triton Digital, 2015. 3: NLO, 2015. 4: Spotify, 2015. 5: RAJAR, 2015. 6: RAJAR, 2015. See page 39 for further information FM/AM DAB DTV Online

Page 10 Page 11 es of terrestrial radio are well suited to measure- publishers and analysis of the devices being used to the whole population in order to deliver adver- in zero ratings for smaller stations at certain times ment using samples, and the currencies on which to access audio content. tising currency figures. This approach is statisti- of the day. they are traded are long established, accepted by cally very robust for estimating the audience of When measuring online usage, where panel meth- media buyers and employ quite simple audience / / Two approaches to measuring terrestrial radio stations, which are generally small odologies are typically used and consumption is metrics. the audience at the market level: in number and large enough in scale within any usually fragmented, even large panels face limi- given measurement area to deliver stable audi- With the exception of a few large publishers, such samples and census tations in accurately measuring smaller and more ence figures. The limitations of sample methods in as Spotify, Apple Music, Deezer and Pandora, targeted audiences. Similarly, panels quickly lose The use of sample-based methodologies, such as this regard become more apparent when passive most online audio channels have tiny audiences accuracy when trying to measure audiences with day-after recall and panels, allow estimations of devices are used: for example, as the cost of an compared to those of terrestrial radio stations. multiple attributes, as the number of panellists po- reach and time spent consuming media, and for electronic measurement panel is high compared to Therefore, sample-based measurement is unlikely tentially matching all of a series of compounding basic demographic information to be attributed. declarative diary panels or interviews, they tend to to provide a robust basis for a reliable currency on criteria results in a small sample that cannot reli- Census data is (theoretically) a comprehensive and be small in relation to the total population, and if which advertising can be traded. Audience sizes for ably be extrapolated to the total population. exact account of all measurable consumption be- the unit of reporting is too granular, this can result these publishers can be inferred from census-level haviour for all selected publishers/channels, but it measurement of individual streaming sessions, is a challenge to link that consumption data with but these techniques cannot be applied to terres- figure 3: demographic or other non-personalised informa- trial radio. tion to build detailed profiles of the consumers. An overview of the capabilities of sample and Chapter 3 / / Sample measurement census data Overview of audience Sample measurement, in which a statistically rep- resentative sub-set of the population is identified Sample data Census data measurement and recruited to participate, forms the basis of (User centric) (Site centric) methodologies traditional television and radio audience measure- ment. Radio measurement in most countries is Respondent level data reflective Comprehensive and exact In broad terms, measurement and analytical meth- based on declarative techniques, predominantly of users’ behaviour account of all online activity odologies fall into two types, the first of which is by computer-assisted telephone interviewing restricted to a single publisher’s audience with the (CATI) or in a smaller number of markets by diaries Capabilities: second offering insights across a market, includ- (paper or electronic, or a combination of the two). • Reach estimates (unduplicated) • Total usage (contacts, user interactions, etc.) ing several publishers. On the one hand, publish- Some countries, such as the US, Norway, Denmark, ers can leverage a range of tools and services to Sweden and Switzerland, amongst others, employ • Frequency estimates (unduplicated) • Technical information about all measured analyse streaming of their own content, and these electronic devices such as the Personal People streams (e.g. device type, operating system, IP • Demographic profiles data can be used for increasingly sophisticated tar- Meter (PPM) or Mediawatch to allow passive au- address, time of day, etc.) geted advertising, based on location, time of day, dience measurement, which does not rely on the • Cross-platform behaviour at household/ device type, etc. Several companies offer compre- listener recalling his or her past listening. • Accurate measurement of small and niche pub- individual level hensive suites of tools to deliver audio to listeners lishers and content Not only do samples offer the only viable solution and analyse performance in real time, and tech- • Replicates of traditional RAM metrics for measuring terrestrial radio, given the lack of a nologies of this type form the basis of the nascent return path for most radio devices, they also have automated and programmatic platforms that are Limitations: the advantage of attributing demographic and oth- starting to emerge. er information to the listeners, allowing an analysis • Accuracy decreases for small and niche content • No accurate estimate of unduplicated reach or On the other hand, independent, market-level re- of both what has been listened to and by whom. frequency • Potential bias due to sample composition (e.g. search potentially offers a more holistic view of the However, samples provide only a snapshot of the for online panels, may be skewed towards • No demographic or other forms of information market, allowing comparisons between different total listening, and the data must be extrapolated heavy Internet users about users

Page 12 Page 13 / / Alternative tracking methods Probabilistic identification relies on analysis of / / Census-level measurement for the mobile environment: thousands of anonymous (non-personally iden- tifiable information) data points to recognise a Census-level data, sometimes referred to as cross-device identification unique user across multiple devices, employing al- machine, technical or return-path data (RPD), gorithms to determine the probability that a par- theoretically delivers an accurate account of to- In response to the decreasing accuracy and rel- ticular user is matched to one or more devices. The tal consumption, potentially across all devices. evance of cookies in the multi-device marketing parameters used in probabilistic device recognition It should be noted that whilst it may be techni- environment, a number of approaches have been include: screen resolution, device type, operating cally possible to measure each stream, includ- developed. To date, there is not a universal solu- system, IP address, time of day, environment, and ing what, where, when and on which device it tion that can be employed to address cookies’ lack behavioural/browsing data, among others. Proba- was delivered, there are a number of reasons of relevance when measuring or tracking mobile bilistic ID vendors, such as Drawbridge and Tapad, why in practice this may not actually be the audiences, and many advertisers and publishers state that they can match devices with about 70% case. are currently using a combination of approaches to 91% accuracy4. to maximise data accuracy2. The major players in Census measurement is typically carried out by the online industry – Google, Facebook, Microsoft, the insertion of tags, or tracking pixels, into web / / The current state of user Unlike deterministic identification, the probabi- Apple and Twitter – have all developed technolo- listic approach is not restrained by closed eco- based content assets such as websites, and tracking: from cookies to cross- gies that allow users to be identified and tracked, systems, meaning that advertisers may be able cookies are still an important measurement device identification opening the door to targeting capabilities that al- to identify and reach anonymous users across technology. In recent years, leading analytics low advertisers to reach people as they switch be- multiple platforms and properties. However, this companies such as Nielsen, comScore, Quant- One challenge has been overcoming the limita- tween desktop and mobile devices. approach is less mature, and vendors’ methodolo- cast and Google have developed SDKs (Soft- tions of cookies as a technology for measuring gies are closely guarded secrets, making unified ware Development Kits) and APIs (Application Internet usage. Cookies, which are small pieces There are two basic approaches to cross-device auditing across solutions difficult. Programme Interfaces) that allow publishers tracking, or device matching: deterministic and of data stored by a user’s browser when visiting to integrate measurement capabilities that can probabilistic. The deterministic approach identi- a site, are no longer a reliable proxy for humans. deliver unduplicated audience data across mo- fies users across multiple devices when they are Before the emergence of smartphones and mo- bile platforms and other connected devices. bile tablets, and the decrease in cost in personal logged onto platforms such as email or social me- computing devices, it was reasonable for mea- dia accounts, and it therefore uses first-party Per- Whilst the elimination of the need to extrapo- surement purposes to approximate a cookie to an sonally Identifiable Information(PII). Deterministic late data to the wider population is advanta- individual, as most only used a single device and a identification requires scale, and more specifically geous compared to the limitations of data single browser. This is no longer the case, as usage access to a large base of logged in users, and it can derived from samples, census-level data gives has fragmented across multiple devices. Cookies therefore only be leveraged by the giants of the information about what was consumed. It says 3 are often blocked or deleted by users, reducing industry. It is highly accurate (95%+) , and very ef- little about the characteristics of the people ac- their ability to deliver reliable audience data. fective for executing advertising campaigns within tually listening to the audio content. For a com- the walled garden ecosystems of, for example plete understanding of who, as well as what, Furthermore, cookies are less effective on mobile Google or Facebook. However, the use of PII raises was consumed, census-level data derived from devices than on desktop browsers: some mobile privacy concerns, and deterministic identification audio streams needs to work alongside a panel web browsers do not accept third-party cookies, does not allow users to be tracked beyond these or other sample-based methodology that can which are typically served alongside display ad- aforementioned walled gardens. deliver information on demographics, such as vertising, whilst the walled garden architecture of age, gender and other attributes. mobile apps prevents cookies from being shared The leading examples of deterministic tracking between apps that are used to access online con- include the Google Advertising ID, (which replaced tent. Google’s previous Android Advertising ID), Apple’s ID for Advertisers (IDFA, which replaced the company’s earlier UDID) and Facebook’s user ID (UID).

Page 14 Page 15 online and offline listening, and this is conducted low cost, as no additional measurement technolo- stripped out and replaced by instream spots or Chapter 4 by a number of European JICs, although it is by gies or methodologies are required, this approach other radio content. However, in markets where no means universal. A more targeted approach to has the limitations of any purely sample-based the RAM – and therefore the trading currency – Strategies for understanding listening patterns is the platform measurement system in that it does not provide does not differentiate between the different devic- measuring audio usage report, examples of which are published in an account of overall consumption. As with any de- es on which radio is consumed (FM, DAB, IP, digital the US, UK, France and the Netherlands; however, clarative measurement, it relies on the accuracy of TV, etc.), it is not possible to monetise simulcast audiences these reports do not offer the granularity required participants’ memories, both in terms of what they adbreaks separately, as they form part of the sta- for media planning and buying. More detailed mar- were listening to and via what devices, whereas tion’s total reach and therefore can only be sold as At the most basic level, publishers may use their ket-level data can be found in digital audio brand the online world is largely measured by passive part of the terrestrial spot campaign. own first-party streaming data to sell their online usage reports, which typically use census-level and therefore theoretically much more accurate audio inventory, if indeed they choose to monetise measurement to quantify online audio channels by means. / / Platform usage reports these additional impressions, thereby circumnavi- stream starts, listening durations and other usage gating the need for third-party audience measure- Reliance on traditional RAM is also likely to miss The prevalence of online audio listening, as well as metrics. These reports are only available in a small ment. Increasingly, this data may form the basis for significant parts of the online audio landscape. As use of other forms of audio distribution, is reported number of markets, and they are delivered by a va- targeting online audio spots, for instance by geo- they are primarily designed to measure a relatively in a number of markets through studies that anal- riety of organisations, from official measurement location, or be used in programmatic trading. small number of high reach – at least locally – ra- yse platform, device and service usage. In some providers to commercial companies. dio stations, RAM samples are not well suited to cases, these are carried out by the national radio The advantages of this approach are relatively The most technically ambitious approaches to measuring the long tail, which may include broad- audience JIC, in others they are released by private low cost, immediate availability of data and au- measuring online and offline audiences together casters’ digital brand extensions and online-only research organisations. tonomy, in that the publisher does not need to in order to produce a convergent currency take radios. Digital pure players, such as Spotify, Deezer find agreement with any other actors on the mar- The NLO in the Netherlands, for exam- the form of hybrid measurement methodologies, and Pandora (in the US) are not part of existing ra- ket. However, the disadvantages include potential ple, releases its Audio Distributieonder- which combine traditional sample-based and dio audience surveys, although in some countries non-standardisation of data and reporting across zoek once every two years, using a com- census-level techniques. Hybrid measurement for there are discussions over whether such publish- the market, which may lead to confusion among bination of an online questionnaire (CAWI) and a radio/audio is in its infancy, and early examples can ers should be included in radio measurement. buyers, and the absence of a third-party auditing single day’s worth of data from the radio diary that be found in the sections on France (page 24), Ger- or accreditation organisation that can validate the Furthermore, traditional RAM methodologies do is used for regular RAM. The most recent edition, many (page 26) and the US (page 30). publisher’s figures. Furthermore, the data is likely not offer an opportunity to leverage the richer data released in September 2015, was based on 4,966 to offer insights only into stream starts and dura- / / The single-source approach to that may be applicable to radio stations’ online au- responses, and it offers a comprehensive analysis tions, and unless the publisher has some form of of radio listening platforms. The report also in- measuring terrestrial and online dio inventory. Where FM and online audio streams registration, no demographic information can be are measured holistically, these are effectively cludes some data on online audio music services. appended. This latter point may be mitigated by radio audiences treated as equal in terms of their targeting possi- RAJAR, the UK’s radio measurement JIC, inferring audience profiles either by reference to bilities and are therefore valued at the same CPM. The simplest approach to measuring listening to releases a platform usage report twice a radio station’s offline audience (in the case of It is only by trading outside of the RAM-based cur- radio brands across different platforms is to adapt a year under the name MIDAS (Measure- simulcasts) or through the use of third-party de- rency that publishers can enrich their online audio the existing RAM methodology to differentiate be- ment of Internet Delivered Audio Services). This sur- mographic matching solutions, such as Experian’s inventory with additional data, deploy retargeting tween listening on FM, DAB+, online, etc. RAJAR vey is conducted using an online diary that is quite Simmons National Consumer Study. and other more sophisticated audience segmenta- in the UK, for example, includes this as a question similar to that used for RAM in the UK, and the tion and thereby sell at higher CPMs. However, to go beyond this reliance on first-party within the diary that panel participants are asked sample of approximately 2,000 demographically data, a number of third-party audience measure- to complete, which allows the organisation to By trading online audio inventory outside of the representative respondents are individuals that ment and analytics methodologies have been de- break down listening per station by platform. Simi- single-source RAM, radio stations have the flex- have in the past completed the regular listening veloped and are available on different markets. larly, AIMC in Spain, using a combination of CATI & ibility to introduce non-linear components to ad diary. MIDAS allows a rich and detailed snapshot CAPI methodologies, can differentiate by platform breaks in their simulcasts. As is common in the of the population’s audio behaviour. In addition to The least complex model is to measure the audi- for stations that are members of the organisation. US in particular, and to some extent in the UK, the audio time spent listening, RAJAR collects data on ence holistically by adapting the existing single- advertising played out on terrestrial radio may be activities being undertaken whilst listening, a wider source RAM methodology to differentiate between Whilst this has the advantages of simplicity and

Page 16 Page 17 repertoire of listening locations than is contained About 2,400 people from across Switzerland were Spotify released a report titled The New reports do not contain data on individual channels, within the core radio survey and other supplemen- surveyed by telephone and online for this study. Audio in September 2015, using TNS although they may offer some insights into the au- tary information, but the data is not broken out by online panels consisting of more than dience size of the few largest actors on the market, In the US, Edison Research and Triton individual stations or publishers. 20,000 respondents in 10 European markets. The for example Spotify and Pandora. Digital publish The Infinite Dial. This tele- research sought to compare the reach of Spotify’s An accurate picture of radio listening share by phone interview study of approximately Platform usage reports offer good insights for free (advertising supported) service to radio sta- platform is particularly significant in the UK, as 2,000 Americans has been running since 1998, the market they measure, but they do not allow tions in each country, as well as to the reach of one of the preconditions that must be met before offering detailed insights into digital audio con- an accurate assessment of online audio listening other audio streaming services, and the published the planned migration from analogue to digital sumption, attitudes to media and technology and beyond those borders. The methodologies vary report heavily suggests that Spotify adds incre- broadcasting (DAB) can begin is that 50% of radio longer-term behavioural trends. The 2015 edition widely between those few markets that do have mental reach to radio stations. Whilst interesting, listening is carried out via digital platforms. Digital found that 53% of Americans (12+) listen to on- an established survey of different listening plat- The New Audio is by no means an independent re- listening includes DAB, online/apps and DTV (digi- line radio each month, and that in terms of online forms, as does the scope of what publishers are search study. tal terrestrial television). In October 2015, RAJAR audio brands, Pandora is the clear market leader, included, and results cannot reliably be extrapo- reported that analogue listening had fallen to 50%, followed iHeartRadio, iTunes Radio, Spotify and / / Advantages and limitations of lated from one market to another. For this reason, with digital standing at 41.9% (the difference being Amazon Music. Triton Digital’s Webcast Metrics platform usage reports it is so far difficult to really understand the impact listening that could not be attributed to either ana- Top 20 Domestic Ranker (which includes only data on radio listening of international services such as logue or digital platforms). More than two thirds from participating publishers) gives an insight into Reports such as RAJAR’s MIDAS and Edison Re- Spotify, iTunes Radio or Deezer across the many of digital listening is through DAB radio sets, with the performance of both pure players and radio search/Triton Digital’s The Infinite Dial offer a valu- radio markets in which they operate. much smaller proportions for online and DTV lis- broadcasters, with Pandora, Spotify and iHeart- able, helicopter view of online audio listening. With tening. Radio leading in terms of average active sessions, some of these reports, it is possible to understand / / Digital audio brand usage followed by Cumulus, CBS Radio, NPR member the scale and compare consumption levels on dif- reports In France, Médiamétrie publishes a stations and ESPN Radio. ferent devices and platforms, such as broadcast number of different reports that con- Ongoing, regular ranking of digital audio radio and online audio, and when conducted over a tain insights on cross-platform listen- Edison Research launched Share of Ear in 2014, services and in-depth market analyses period of time, it is possible to analyse trends and ing behaviour. Most recently published in October and announced that the report’s delivery schedule exist in a small number of markets. In evolving patterns of behaviour. They typically use 2015, Radio 2.0 looks at online radio and music would increase from two to four times per year due Germany, Goldmedia publishes an annual report well established and trusted research methodolo- consumption, using data from the Mediametrie// to industry demand. Share of Ear is a single-source that gives an overview of providers, content, use, gies, predominantly diaries, questionnaires and in- NetRatings Internet measurement and the organ- measurement of Americans’ share of time spent advertising revenues and forecasts. Goldmedia’s terviews, and are carried out by respected research isation’s Media In Life study. Médiamétrie has in listening to all audio sources, including broadcast Webradiomonitor has been running since 2009, organisations. the past also released data on different platforms radio, Internet-only streaming audio, podcasting, and it is based on survey data from all online audio based on its regular audience measurement stud- satellite radio, TV music channels, and listeners’ However, these sample-based reports are typically providers in Germany. ies and ad hoc research. own music collections. The study uses paper and low on granularity, giving overall listening by plat- The German market also has regular reporting of online diaries in English and Spanish that record form and giving limited data on individual stations In Switzerland, the DigiMig working the online audio channels in the country. Published audio listening of a 24-hour period, using a sample or publishers. This lack of granularity limits their group – which is composed of the radio by the JIC agma, ma IP Audio gives the performance of about 2,000 individuals. The full data is available relevance for audio advertising currency purposes, industry and regulators and aims to as- of more than 300 channels from 57 publishers, to subscribers only, and Edison publishes selected as they usually cannot be used for media planning sist the transition to digital radio – has published a ranging from simulcasts of FM stations to online data points from the study. The May 2014 report and buying. They may be published regularly or on pilot study on digital radio listening. The pilot Digi- only and user generated radios. Ma IP Audio is a showed AM/FM radio with a 52.1% share of time a more ad hoc basis, and the frequency of reporting tale Radionutzung report was released in August census-based audience measurement, using the spent listening to audio, followed by owned music is usually low at a few times a year or just once ev- 2015, and its headline findings indicated that 45% log files generated by participating stations. For at 20.3%, Internet radio/music at 11.6%, the satel- ery two years. Furthermore, these reports depend of all radio listening in Switzerland is now digital. more information about online audio measure- lite radio service SiriusXM at 7.7%, TV music chan- on declarative methodologies using small samples, DAB+ and online were found to be equal in share. ment in Germany, please see page 26. nels at 5.2% and podcasts at 1.7%. which is in stark contrast to the passive, continu- The study was undertaken by GfK, and there are ous and big data methodologies used to measure plans for two surveys of digital listening per year. other digital media consumption. Typically, such

Page 18 Page 19 In France, a partial ranking of online / / An example of audiovisual eo advertising is carried out according to the IAB radio channels is published by ACPM hybrid audience measurement VAST (Video Ad Serving Template) standard, which (formerly OJD), the organisation that is essentially creates measurement events. Publish- responsible primarily for certifying the circulation in development: SKO Video ers insert Kantar Media tracking pixels into the of newspapers and periodicals in the country for Data Integration Model (SKO- video campaigns that they sell direct to advertis- the purpose of audience measurement for ad- VIM) in the Netherlands ers (as opposed to through an automated trading vertising. Numbers of listeners and total listening platform), and when the commercials are served, hours, both in France and worldwide, are included, A number of TAM organisations are currently these pixels place a call to Kantar Media’s sys- and a ranking by publisher brand, group/network developing hybrid solutions to measure view- tems. This data, which also identifies the type of and individual channel is released on a monthly ba- ing of television content via traditional means device on which the spot was played, is therefore sis. The reports do not include all of France’s radio alongside online video consumption. SKO – the independent of the publisher’s ad server. broadcasters, and the online streams of Lagardère Dutch television JIC – is among the most ad- The census-level data is derived from tagging/li- and RTL – two of the country’s largest broadcast- vanced in this field, and in 2016 the organisa- braries inserted into content by participating pub- ers – are not included. For more information about tion became the first in the world to publish viewing moment reporting gives the performance lishers. Identifiers within this census data identify online audio measurement in France, please see rich and detailed online viewing ratings as de- of a publisher over a period of time, rather than the members within the online panel, allowing demo- page 24. livered by a hybrid measurement methodology. previous paradigm, by which reporting related to The next phase, expected to be delivered later graphic data to be appended to the census data. the actual time a piece of content was broadcast. In the US, Triton Digital’s Webcast Metrics in 2016, will see SKO publish the total reach of This process requires input from the participating The move away from a measurement environ- ranks top performing online audio pub- programmes and video commercials as viewed broadcasters, as they have to insert the tags and ment in which there are a relatively small number lishers. Triton Digital’s Domestic Ranker over broadcast and via online platforms. libraries into their players and content in order to of broadcasters towards one in which a larger and shows streaming that was carried out within the be measured. more unstable group of publishers – who may en- SKO’s hybrid model consists of three main com- US, whilst its All Streams ranker gives the quantity ter and leave the measurement over time – means ponents: an online video panel, census mea- of streams without verifying where they originated / / Moving towards an integrated that audience shares will no longer be published surement, and the existing TAM panel (which from. For more information about Webcast Metrics, measurement within the viewing moments report, as these will uses fixed location People Meters). please see page 30. become less relevant for media planners. From At the time of writing this report, the online panel January 2016, cross-platform campaign reporting / / Measuring the online video has been established, and hybrid reporting has / / Continuous, panel-based became available. online audience measurement audience: panel + census commenced for programmes, but there remains some work in the extension of the census project No agreement has been reached to date regarding Two approaches have been developed for panel- The new online household-level panel, oper- to cover all broadcasters as well as all online-only a hybrid currency that will cover media planning, based online audience measurement: firstly, soft- ated by TNS/Kantar Media, provides a single video content. A number of important questions trading and billing. For television, the currency is ware meters, which are locally installed on each source for online behaviour across all online still need to be addressed, such as how to deal Live +6 days viewing whilst for online video there device and may be adapted for different pieces devices. Panellists log onto their connected with programmatic campaigns. Calculation and will probably start to be a discussion on the mar- of equipment, such as PCs, smartphones, tablets devices at the start of each session by select- reporting rules are available on the SKO website. ket the moment SKO start delivering data on cam- and Smart TVs; and secondly, Wi-Fi routers, which ing their names from a drop-down list, either paigns in spring. The next phase SKO has to deliver is to develop capture Internet streaming data from all devices through a home page login for desktop or laptop a combined measurement model that will deliver within the household to which they are connected. computers, or via an app on mobile devices. In the case of mobile devices, this process is ac- viewing figures for programmes across online and Whilst software meters are cheap to install, as tivated when a relevant app, such as a browser offline platforms. Currently, reporting of televi- no hardware is required in the panellist’s home, or video platform, is opened, and this measure- sion and online video viewing of TV programmes well suited to providing very accurate and rich ment software is available for Android and iOS remains separate, but broadcasters can analyse data including time spent watching or listening platforms. the two types of viewing alongside each other to to streamed media, they do place some burden better understand cross-platform audience char- on panellists, especially if the meter has to be The measurement and reporting of online vid- acteristics. It is important to note that the new

Page 20 Page 21 installed on several devices that they use. Some forms the basis of the emerging hybrid measure- is currently being refined in consultation with the software meters, such as that used in comScore’s ment methodologies that are being developed for industry (see page 26); in France, Médiamétrie is Chapter 6 Unified Digital Measurement, allow the different us- Internet, television, radio and cross-platform con- developing a hybrid methodology for audio in par- ers of a particular device within the panel to be sumption. allel with a system to capture all video viewing, and A look ahead identified without the need to log in each time, work is currently underway to ensure all listening Significant evolution of radio audience and this is useful in reducing panel fatigue (and / / Hybrid measurement for is measured, regardless of whether the content is measurement to incorporate online audio is the turnover that results from this and potential accessed by publishers’ own audio players or via Internet audiences expected to take place during the course of 2016 biases6. Wi-Fi routers, on the other hand, reduce third-party aggregators (see page 24). for the markets featured in this report: France, the burden on panellists, and they eliminate some Hybrid Internet audience measurement solutions Germany and the US. In all three markets, a of the technical challenges associated with mul- have been in place for some time, using a combina- / / Obtaining the agreement of number of issues remain to be resolved, both in tiple devices being used within a household. This tion of census-level data and panels. Online panels all publishers and platforms terms of technical capabilities and agreement greater usability comes at the cost of measure- are relatively cheap to set up and run, especially between the various actors on these markets. ment precision, which is lower with routers than compared to panels that use electronic equipment Census-level measurement of audio streaming, Beyond the purely methodological challenges it is with software meters They can only measure such as the PPM, as they only require a software which is a key component of any hybrid measure- arising from extending traditional RAM into the in-home usage, and in some cases they can mea- install on the participants’ devices. This allows ment solution, requires the participation of pub- online environment, more fundamental questions sure access to pages using the HTTP protocol but measurement organisations to achieve sufficient lishers and online radio aggregators to deliver a will need to be addressed in several countries, not the secure HTTPS protocol, which is used by scale to measure large numbers of publishers and comprehensive account of all listening. The experi- such as whether to include online audio pure roughly a quarter of the Internet’s most popular websites. ence of France and Germany suggests that obtain- players into the RAM and how advertising trading websites, including YouTube. ing the agreement of radio broadcasters to imple- / / Hybrid measurement for ment the necessary tracking technology in their currencies may be best adapted in the future. To date, the first use of these digital measurement television/video audiences audio streaming players is not particularly difficult, However, as the consumption of online audio – techniques has been seen in the next generation of but that covering listeners that tune in via aggre- from both terrestrial radio broadcasters and online television audience measurement (TAM) projects Television audience measurement organisations in gator platforms can take longer to achieve. only publishers – continues to grow, and digital that are being developed to unify online and of- the US and several European markets are currently marketing becomes ever more enriched by sophis- fline TV viewing in convergent, or at least compa- developing solutions to measure the TV audience Compliance on the part of broadcasters is also ticated data collection, management and targeting rable, currencies. Whilst they may theoretically be across all devices, including traditional broadcast something that has slowed the development of capabilities, significant advances are likely to be applied to RAM in the future, the use of software and connected devices, both live and on-demand. hybrid television/online video measurement proj- seen beyond the market-level measured audience. meter and Wi-Fi router solutions are not currently For an overview of recent developments, please ects, which bear some similarities to hybrid audio European radio has taken its first steps into pro- being used in market-level audio measurement (in see egta’s document Advances in hybrid television measurement. For example, BARB, the JIC respon- grammatic audio trading, and programmatic solu- isolation from other online media), with the excep- audience measurement, which is available to down- sible for television audience ratings in the UK, is tions are starting to be deployed by terrestrial ra- tion of Nielsen’s SDK (Software Development Kit) load at http://bit.ly/egtaTVmeasurement02. rolling out the first stage of its Project Dovetail, dio sellers in the US, potentially negating the need in the US. which delivers market-level data on television / / Hybrid measurement for broadcasters’ online video viewing for the first to overhaul existing RAM technologies and meth- Chapter 5 radio/audio audiences time, in beta format, as not all of the country’s odologies altogether. broadcasters have implemented the tracking code The hybrid audio measurement methodologies The radio industry made its first steps towards in all of their video players. In part, this is a con- currently being adopted have the advantage of Hybrid measurement: hybrid audience measurement in 2014 and 2015, sequence of the numerous operating systems and bringing some elements of digital accuracy and but as of early 2016 there are no fully operational merging terrestrial devices that these players have to be adapted for, accountability to those markets where they are hybrid solutions being used to deliver audio trading each of which is likely to undergo regular software being developed, but a faster pace of change is to and online audio currencies. In the US, Nielsen has been developing updates, and in part it reflects the challenges of be expected in technologies used to combine pub- a unified measurement solution, which has yet to measurement and working with broadcasters’ technical teams who lishers’ first-party data with third-party data and come to market (see page 30); in Germany, the JIC have not previously been involved in audience deeper insights about individual users in order to metrics agma released the report from its hybrid ma Audio measurement issues. enrich online audio inventory for impression-based methodology in December 2015, and the system The combination of sample and census-level data – rather than audience-based – transactions.

Page 22 Page 23 currency for Internet in France on every screen measurement solution, combining census-level sive account of the French market, as the stations (desktop/laptop, mobile, tablet and Internet over- and panel data for online audio consumption. The of some major broadcasters are not included in the all). Along with this user-centric measurement, organisation is currently working towards a con- published figures. eStat is a site-centric, census-based measurement sensus on the market in order to produce an audi- The figures in ACPM’s web radio ranking are derived platform, which uses tags inserted into participat- ence platform dedicated to audio. from two sources: participating publishers deliver ing websites and video players. eStat also offers France Médiamétrie is trialling its own portable electronic streaming data from their own web analytics tools, mobile app analytics. meter, called RateOnAir. This device uses water- and this data is verified and audited against raw Radio audience measurement in France is carried The hybrid solution currently under development marking to measure radio audiences, in much the log file data extracted from CDNs (Content Delivery out by the research agency Médiamétrie, which is for audio in France can best be understood by ex- same way as the Portable People Meter developed Networks, interconnected systems of cache serv- also responsible for measuring television, Internet amining the work Médiamétrie is putting in place by Arbitron (now Nielsen Audio). This electronic ers that use geographical proximity as a criterion and cinema audiences. The main methodology is for video. The television broadcasters and online- measurement, which is complementary to the cur- for delivering Web content). Whilst ACPM collects day-after recall CATI, and the measurement is pub- only services, such as YouTube and Dailymotion, rency RAM, is expected to play an important role in data on all audio streaming activity, only the listen- lished under the name 126 000, due to the number have implemented the eStat measurement tags the future measurement of online audio in France. ing figures of subscribing publishers are included of respondents in the survey each year. The radio into their players, which provide the census-level Using RateOnAir’s encoding (watermarking) tech- within the ranking. audience figures are published four times per year, usage data. This is combined with Médiamétrie’s nology, broadcasters can differentiate between and a supplementary three-week diary, called the The ACPM reports were initially developed to pro- panels, which qualify the video usage and at- panellists listening live to their FM and online audio Panel Radio, delivers insights on audience behav- vide authenticated quantification of the digital tribute demographic profiles. This research has streams, and the device may potentially be used iour and data that can be used for media planning audiences of music radio stations and web radios, been carried out for about three years and reports to identify and measure on-demand podcasts and purposes. and several important audio companies are not in- monthly. The hybrid methodology has two main other time-shifted listening. cluded. Firstly, France’s predominantly talk-based Radio stations in France sell their whole audi- principles. First, the census usage information col- There remains a discussion within the French radio stations, such as Europe1 and RTL, do not sub- ences, including every device, every platform and lected across several websites is used in the global market over the future scope of the measurement, scribe and are therefore not included. Secondly, every location for which listening takes place. To weighting process of the panel in order to achieve and specifically whether to include pure players from their inception, the reports have contained have access to audience figures on device usage, conformity between the two measurements. such as Spotify and Deezer. These services are only live, passive online audio listening, hence ex- radio stations currently have to subscribe to a spe- Second, the usage calibration (video viewed, time currently included within the existing Médiamétrie// cluding on-demand services such as Spotify and cific survey, called Global Radio, published once per spent) is based directly on the census measure- NetRatings Internet audience measurement. Deezer. year. They are able to report this information to the ment. market, although it cannot be used for media plan- According to ACPM, whilst the issue of fraudu- A truly comprehensive television/video measure- / / The existing web radio ning. lent and other non-human activity is gaining im- ment is expected to be launched during 2016, cov- measurement report produced portance, it is not considered to be a major issue In 2016, all subscribers to the 126 000 will have ering four screens – TV, mobile, desktop and tablet. by ACPM for online audio streaming in France at this stage. access to a breakdown of delivery by device, and Methodologically, there is little difficulty in mea- ACPM does have a control in place, as the organ- this will be extended to two publications (but still Since 2013, ACPM (formerly OJD) – the body that suring total radio/audio audiences across all de- isation has constructed a database of IP addresses without media planning). audits circulation data of the press in France, and vices using the same techniques. However, the that it can use to filter out suspicious traffic. more recently Internet traffic data – has published main challenge lies in the different models by / / Technology in place for a listening figures for online radio channels in the The ACPM figures do not constitute an advertis- which television and radio content are accessed country, from both FM broadcasters and online- ing currency, as online audio is primarily transacted hybrid RAM solution online. A part of online radio listening, unlike on- only radio stations and groups. A monthly press re- on impressions or GRPs. It serves as a valuable line television viewing, takes place not through the Médiamétrie’s capabilities with regards to mea- lease gives information about the number of active source of additional, third-party verified audience stations’ own players but via aggregator services, suring online audiovisual content reside with its listening sessions of French radios (in France and data that publishers can use in their marketing and and there is therefore a larger and more diverse set Internet department, which is responsible for for the whole world), total listening hours (whole sales argumentation. It is also a useful tool for me- of market actors that need to implement the eStat measuring Internet audiences in France, combin- world) and average listening duration (whole world) dia agencies, as it provides insights on the market tracking technology in their players. Médiamétrie ing two platforms: Médiamétrie//NetRatings and for each publisher and group/network of channels. and can inform their media buying decisions. has the technology in place to achieve this hybrid eStat. Médiamétrie//NetRatings panels provide the However, this service does not give a comprehen-

Page 24 Page 25 ers. The data published in that report showed that dio, and a new and supplementary diary survey of digital data collection into the scope of Germa- about one fifth of Germans aged 10 and over lis- of about 3,000 participants, which will bridge the ny’s radio audience measurement, allowing audio tens to online audio, and it has a daily reach of just gap between the online audio listeners and their sales houses to identify not only their broadcast under 7% of the population. ma IP Audio is a first social-demographic profiles. Respondents indicate reach but also the incremental reach delivered by step towards a convergent currency that would which stations they are listening to during hourly IP-enabled devices, offering metrics in which the Germany meet the demand from media buyers for inde- slots throughout the day, as well as their daily ac- market can be confident whilst offering flexibility pendent and robust data and further develop the tivities (e.g. asleep, at home, at work, etc.) and the over the sales models that can be applied to each Radio in Germany is measured by the day-after channel as an advertising platform. device they use for each station and hour of listen- portion of the inventory: online audio may be sold recall (CATI) method, with data published in two ing. agma has developed a single source approach, alongside broadcast as a traditional radio product, This research is based on log files generated by waves per year under the name ma Radio. Stand- whereby online diaries are completed for 14 days or it may be identified separately and sold under streamed audio sessions. These include the online ing for Media-Analyse, this study is undertaken by alongside the traditional ma Radio CATI survey us- a digital paradigm, such as programmatic trading. simulcasts of radio broadcasters as well as other the JIC agma, with fieldwork carried out by a num- ing the same respondents, with additional respon- forms of audio, such as web-only channels, radio The first test and proof of concept following the ber of research organisations. This methodology dents recruited via an online access panel. Under aggregators and user generated radio services publication of ma 2015 Radio I and ma IP Audio I in provides an accurate and reliable audience figures this data modelling and fusion model, personalised such as radionomy and laut.fm. 310 channels from mid 2015 led to a period of further development of for the German radio market, which is a highly data from the diary survey will be projected onto 57 publishers plus two user-generated radio sta- the data handling and testing the chosen model, fragmented landscape featuring some 400 radio the log file data, allowing for a combined reach fig- tions were reported in the ma IP Audio reporting and the first publication of ma Audio took place in stations in the public and private radio sectors, the ure of broadcast and streaming for each participat- of September 2015, representing services that December 2015 based on ma 2015 Radio II, ma vast majority of which are available on a regional ing station. both carry advertising and those that do not. The 2015 IP Audio II and the diary. basis only (disregarding online distribution). The number of publishers sharing ma IP Audio is still The goal is to create comparable performance launch of digital terrestrial broadcasting (DAB+) in increasing. In December 2015, Spotify data were values for programming and advertising across / / Webradiomonitor: a source of 2011 introduced national stations to Germany for published in the reporting for the first time. For broadcast and online channels, as follows: the first time; however, digital radio receiver pene- further insights on digital audio 2016, more than 400 channels will participate in tration remains low – at 11% of households by De- • Widest audience (users in the last two weeks) listening in Germany ma IP Audio. cember 2015, according to the World DAB Forum • Weekly reach It should be noted that agma, the official audience – and regional/local FM therefore remains by far ma IP Audio currently reports on a quarterly basis, Monday to Friday/Saturday/Sunday • Daily reach measurement body, is not the only organisation to the dominant listening platform and station type giving average sessions per week, two-week pe- The relatively small size of the initial panel natu- deliver reports on online audio usage in Germany. within the German radio audience measurement. riod and month, and average session length per rally presents a challenge to the granularity of data Since 2009, Goldmedia has published a report un- quarter, as well as the percentage change for each In recognition of the increasing importance of on- that can be reported with confidence. The diary der the name Webradiomonitor, which is based on publisher compared to the previous quarter ma- line audio as a means of delivering radio and other is therefore just used for modelling the technical a survey of online audio providers in Germany, in- chine access. Its strength lies in the accuracy of the audio services to listeners, agma introduced a new data. The base of ma Audio is ma Radio, with more terviews with industry experts and desk research. census-level data produced, as this gives an actual report in early 2014, giving what the organisa- than 80,000 interviews. The first ma Audio was This study offers insights into the market players account of the listening activity taking place on the tion refers to as the technical reach, indicating the published in December 2015, and the study will in online audio and their characteristics and for- German market. However, unlike the traditional ma volume of internet audio streams accessed – de- be continued in 2016. The two studies, ma Radio mats, usage and advertising data and forecasts. Radio study, ma IP Audio does not have any mech- rived from census measurement – of participating and ma Audio, are likely to exist in parallel during However, Webradiomonitor is not a comprehensive anism to apply demographic information, such as publishers under the name ma IP Audio. The scale the next phase of development, with ma Radio audience measurement study, in that it does not age or gender, to the listeners behind the streams. of the growth in online audio listening had been continuing to serve as the currency measurement report usage or demographic data for individual established by responses in the earlier ma Radio / / Towards a radio convergent for total reach of each station and the possibility to publishers or channels, and it uses neither declared waves, which demonstrated a 30% increase in re- data from listeners nor directly collected streaming currency in Germany plan these stations by rural district. Greater granu- spondents that had listened to Internet radio at larity may be achieved in the future for ma Audio data as the basis for its analysis. least once (ma 2015 Radio I). The profile of these The next step for the industry is the development by expanding the numbers of respondents to the listeners – typically young and educated, with high of a new survey named ma Audio. This is based supplementary diary,. purchasing power – is of great interest to market- on three pillars – the existing ma Radio, ma IP Au- This new, hybrid methodology brings the accuracy

Page 26 Page 27 nitely more listening due to web radio usage. figure 4: egta: The online advertising world, increasingly also for video and audio, offers the ability to target listen- comparison of media-analyse reports ers with a high degree of accuracy and granularity, including through the use of additional sources such as web browsing activity. Would you expect broad- ma Radio ma IP Audio ma Audio casters to adopt two parallel sales paradigms, either using the published data or on a targeted and auto- Reference population 10+ N/A 14+ mated (programmatic) basis? Type of measurement Sample Census Sample & Census (OL): Yes, definitely. From my point of view, pro- Hybrid, modelling & Olaf Lasalle grammatic has nothing to do with the figures or Methodology DAR (CATI) Log file collection data fusion the reach in the officially published data. I know Reporting frequency 2x per year 4x per year 1x per year in 2015 Managing Director that the agencies want to use programmatic buy- Total and online audio ing, but not because it’s better; it’s only because Type of reach Total Online audio and convergent the agencies have put so much technical infra- FM /AM/DAB+    egta: What is the German market looking for as a structure into these things, now they have to sell Simulcasts    solution to the measurement of broadcast and online it by themselves. But it’s like a black box for the audio? buyers, for the sellers, for everybody, I don’t quite Online-only radio    understand why clients are accepting program- User generated radio    Olaf Lassalle (OL): The main reason behind our matic buying without any quality control, without desire to invent ma Audio was demand from clients Online radio aggregators    any institutions analysing the data. and agencies. They expressed a need to have ap- On-demand audio    Reported publishers Reported proved data for web radio usage. For the past few egta: On that point – with any files generated around Podcasts    years, web radios have been going to the agen- online usage, there’s the potential to record some Demograpics    cies with their reach figures, but the response has non-human behaviour. Do you have some verification Geo-location/region   * been that these data can’t really be used for plan- tools to identify whether the log files derive from real Listening location    ning; they can’t be fed into the planning tools the listeners? Granularity agencies use. For example, in Germany radio has a Target audience    planning tool called RadioXpert, and data from the (OL): Well that’s always a question within market AQH **   research – it’s not a question of true or false, it’s new ma Audio can be used in this tool. Average listeners/hr   *** just a question of whether everyone has agreed on Average listeners/day    egta: And for audio sellers, what would you identify the conditions and KPIs. We have different sources: as the main requirements from the measurement we have the technical reach, which is a true figure Daily reach    system? for the number of streams of a particular station, Weekly reach    and we have a quality control within our system (OL): Definitely they are expecting to see more us- Audience share    that these figures are accurate. And we have the age. That’s the demand. There’s always an argu- TSL    second part, which is the diary, and the combined available Metrics ment between the sellers and the buyers: agen- Av. listening sessions/week    information that we get from these two sources cies point out that listeners to a particular station shows some verification. We’ve taken this model Av. listening sessions/2 weeks    do not differentiate between whether they are to all the market participants, principally the media Av. listening sessions/quarter    using FM radio or listening online via their smart- owners, the agencies and clients, and everybody phones, for instance. But now the sellers can, for has agreed on that as the basis for a currency. You * Less detailed than ma Radio; ** Not published; *** Only for channels in ma Radio. the first time, show that they have a new model are right that the digital people claim we can mea- N.B. This table is a simplified overview of the reporting within the Media-Analyse that can actually demonstrate that there is defi- sure everything – that’s not really true. Page 28 Page 29 Report measures listening on IP-enabled devices, registration data from audio publishers, or the ap- when the company operated under that name, is but it does not distinguish between publishers. The plication of third party station format audience “the average number of persons listening to a par- company’s new solution for digital audio has at its profiling. ticular station for at least five minutes during a 15 heart Nielsen’s Software Development Kit (SDK), minute period.”9 By contrast, Triton Digital builds which allows for census-level measurement of / / Triton Digital’s Webcast its definition of AQH by first taking the Total Lis- US listening via participating publishers’ audio players. Metrics tening Hours (TLH) that a publisher has streamed, The company has been integrating its SDK – which counting only sessions with a duration of at least Webcast Metrics, which achieved MRC accredita- Audio measurement in the United States is domi- serves as a software meter – into the players of five minutes within the reported time period (an tion in 2011, reports audiences based on census- nated by two commercial companies, Nielsen Au- streaming audio publishers, working with tech- increase of the threshold from one to five minutes level data in the form of log files delivered by par- dio and Triton Digital. As of August 2015, Nielsen nology partners such as Triton Digital, Wide Orbit over the THL figure published by Triton Digital), ticipating publishers or its proprietary Application is responsible for measuring broadcast radio au- and the mobile app developer jācapps. According which is then divided by the number of hours in Programming Interface (API), which allows online diences, for which it provides the currency using to Nielsen, in the initial phase the project will be the reported time period with a session duration audio usage on any type of Internet enabled device a panel-based methodology, whilst Triton Digital confined to broadcast radio streams within the 48 greater than five minutes. to be monitored. deploys a system for measuring online audio audi- PPM MDAs, with pureplayers excluded until some Therefore, the Triton Digital AQH is also defined ences using census-level data, under its standard point in the future. This methodology allows Triton Digital to report as “the average number of persons listening to a Webcast Metrics and premium Webcast Metrics Lo- The methodology mirrors Nielsen’s online video online audio audiences using the Average Quarter particular station for at least five minutes during a cal products. Triton Digital is involved in the mea- measurement system, which captures data from Hour (AQH) metric that is predominantly used in 15-minute period.” surement of online audio only, whilst Nielsen has the point of consumption, for example a web traditional radio advertising transactions, as well technologies that cover both broadcast radio and Triton Digital’s AAS is defined as “the average player or mobile app. The SDK detects the station as Average Active Sessions (AAS), which gives a online audio. number of listeners, at any given time, during the identifier, length of listening and timestamps, as more accurate representation of the actual listen- ing audience. The company states that whilst the time period.” AAS is derived by dividing the Total In contrast to most European markets, the US does well as a unique ID number for each listener. Under AQH metric has served broadcast radio’s interests Listening Hours (TLH) by the number of hours in not operate a JIC for audience measurement. Both a partnership, this data is then matched against well, it believes that AAS will ultimately replace the reported time period with a session duration Nielsen Audio and Triton Digital are therefore ac- similar IDs stored by Facebook, which aggregates AQH as the predominant measure of listening au- greater than one minute. Cume is defined by Triton credited by the independent Media Rating Council and anonymises the audience demographic data dience.8 Digital as representing “the sum of unique listen- (MCR). before reporting it back to Nielsen. A further in- ers during the specified time period listening.” ternal calibration panel allows Nielsen to provide The traditional AQH metric, as defined by Abriton, / / Nielsen Audio’s total audience checks and balances. measurement Streaming audience reports will include four basic figure 5: 7 Radio audience measurement in the US is car- metrics : ried out by Nielsen Audio, the company having • Reach: Number of listeners, sessions and quar- An overview of the capabilities of sample and bought out the incumbent service provider, Arbi- ter hours tron, in 2013. The largest 48 designated market census data • Demographics: Gender and age by daypart areas (DMAs), of which there are 273 in total, are • Duration: Time Spent Listening measured using the Portable People Meter, and the • Location: Metropolitan area or DMA Webcast Metrics local remaining markets are measured using diaries. All publishers publishers only National radio ratings, known as RADAR, therefore Nielsen’s SDK solution requires a software integra- combine passive and declarative methodologies. tion on the part of participating publishers, which is • Sessions Started (SS) • Average Quarter Hour (AQH) not necessary within Triton Digital’s Webcast Met- • Total Listening Hours (TLH) • Average Quarter Hour Rating Nielsen has been developing a unified measure- rics, which is discussed below. It is further distin- • Active Sessions (AS) • Time Spent Listening (TSL) ment solution for all audio for some time, and guished from Webcast Metrics in the methodology • Average Active Sessions (AAS) • Cume Rating there has been speculation in the press about an used to apply audience demographic information • Average Time Spent Listening (ATSL) imminent launch. Nielsen’s existing Total Audience to the streaming data: Triton Digital’s solution uses • Cume

Page 30 Page 31 cover broadcast and online has clearly taken some egta: We understand that Nielsen’s objective is to in- time to come to market; what would you say have clude publishers beyond traditional radio broadcast- been the biggest challenges and hurdles to overcome? ers in its unified audience measurement. What is the rationale for this, and do you believe the inclusion of RK: From a technology standpoint, we’ve been non-linear services will benefit the radio – and wider able to leverage much of what was done on the audio – industry? video side and have made tremendous progress on integrating the SDK into the ecosystem. We are RK: Total Audio Measurement will provide the in- now collecting data from more than 2,500 station dustry with a consumer’s total listening behaviour streams. Where things have moved more slowly across devices, platforms and content types. This is around the industry determining what types of data can be used to inform decisions around con- Rob Kass metrics should be reported. Introducing change tent creation, distribution and monetization. Our into the marketplace demands thorough discus- technology is capable of working with linear and Vice President, Digital sion among all of the industry stakeholders and non-linear providers. Ultimately, we believe all Nielsen continues to encourage discussion among parties will benefit from a view of the Total Audi- Audio broadcasters, digital pureplays and agencies. As ence for each platform. those conversations continue, Nielsen continues egta: What sort of trading landscape do you expect to move full-speed ahead in a parallel path: work- egta: Nielsen (and before it, Arbitron) has been de- to see in the future? Will there effectively be two buy- ing with clients, vendors and audio aggregators on livering robust measurement for broadcast radio for ing markets, one trading on audience ratings and the SDK deployment and data collection. years, and Triton Digital’s Webcast Metrics is widely other on impressions, or do you see some way to cre- used to cover online audio. What do you believe the egta: And what are the potential solutions for those ate a more holistic environment for audio advertising? industry is missing right now? challenges and the directions of development you are RK: This is a part of the discussion that the indus- working on now? Rob Kass (RK): Alignment of the disparate met- try is currently having around appropriate metrics. rics for audio content. Just as Nielsen has done RK: Nielsen continues to encourage and facilitate Nielsen is prepared to provide the industry with on the video side, we are developing total audi- discussion among broadcasters, digital pureplays whichever metrics they deem necessary, be that ence for audio. Nielsen’s Digital Audio service is and agencies to align on reporting metrics. average minute audience impression or average the first step in that direction. The Digital Audio quarter hour metrics. egta: Could you take us under the hood of Nielsen’s service was built to align with the radio market- SDK: how does this technology work, what does it al- place: alignment of metropolitan geography and low publishers to do, and how does it allow integra- survey dates, data that combines over-the-air and tion with other technology platforms? simulcast streams, contains estimates of both broadcast and digital content, provides traditional RK: We use the same technology as we do on demographic and daypart breaks and is delivered the video side. There’s a set of code that gets through Nielsen Audio software platforms. The installed by a client, vendor or audio aggregator Nielsen Digital audio service will ultimately include into the browser player or mobile application. The other forms of listening, including on-demand, software development kit or SDK identifies the podcasts and customized audio. In the near fu- content, establishes session duration and listener ture, the industry will have a comprehensive view location. We can also connect listening sessions of a consumer’s listening behaviour, regardless of to other Facebook data using a blind match to pro- device, platform, distribution method or content vide age and gender demographics. And we use type. our PPM panel to calibrate everything to make sure it looks right. egta: The development of a hybrid measurement to

Page 32 Page 33 and they’ll be putting more budget against those and most of those panels were sized to measure al code data on their listeners, so we can target by things, which I think is another reason for the tra- broadcast listening behaviour. And the commercial geo-location. And then certain publishers actually ditional broadcasters to be in the space. If there’s reality is that it’s a digital medium, and buyers ex- have a method for collecting that data and keep- a menu, you want to be on it! pect digital quality measurement. ing it up to date, which has been accredited here in the US by the MRC (Media Rating Council). So, if egta: What sources do you use to collect listening egta: At a practical level, how does a publisher, say a someone lives in New York when they register and data, and how can listening on emerging devices such national radio broadcaster, use your metrics to sell its later they move to Miami, there’s a way to freshen as connected car dashboards, smart TVs and other IP- online audio inventory? that data and keep it relevant. enabled platforms be captured? JR: I think there’s multiple ways. There are several egta: In Europe, it’s rather atypical for broadcasters JR: We require the station or the broadcaster to different metrics that we make available, and the to use a login for their streamed radio, so is there a provide us with log files or use another method to seller can pick and choose the things that are most way to attribute demographic information to these John Rosso actually allow the measurement to happen. There appropriate. The Session Starts metric is probably listeners? are no surveys or panels with what we do, it re- the most commonly used for pre-roll campaigns, President, Market quires a direct connection or a direct relationship as you can forecast how many impressions you JR: Yes, most broadcasters around the world still Development with a broadcaster or publisher. We have two basic are likely to achieve in a given daypart. And the do not gate listening with any sort of login require- methods that we utilise. The first is log files, which other is Average Active Sessions, which is our con- ment, although what we’re seeing is sometimes we collect every day from a publisher. The second currency metric, we can tell you the average num- you’ll have to log in to listen on a mobile device, egta: From your experience, is the selling of online is our API (application programming interface) that ber of listeners you have during a particular time rather than just a desktop or vice-versa. But it’s audio a more significant part of radio broadcasters’ can be implemented directly on mobile devices as period. We also calculate the Average Quarter Hour, typically not 100%, and you’re right, in that case it’s activities in the US than it is in Europe? well as web pages. which is a bit different from AAS because it credits difficult to pull demography out. What we can do, listening to an entire quarter hour if a threshold for some of the reporting that we pass back to the John Rosso (JR): Here in the US, this was really Publishers can use one method or the other, or a five aggregate minutes listening is recorded within publishers, is apply third party audience composi- pioneered by the broadcasters, who were forced combination of both methods. If you just pick one the quarter hour. That follows the way that metric tion curves. Simmons is a great example – they to black out all of the broadcast ads from running method, you don’t capture all listening. We need to has traditionally been calculated in the US. provide audience composition by radio genre, so in streaming due to voice-over talent union work recognise the importance of consumer electronics we can say, for example, if you’re a classic rock rules. This unintentionally created a whole new devices in streaming audio listening today. It’s not egta: So, a publisher can sell online audio like radio or station this is the typical audience demographic tier of inventory that they could fill with ads that just migrating from desktops to mobile. Listening on digital metrics? distribution, and we can apply that. But that’s not were OK to run online. And then, companies like is now happening on all sorts of devices, Smart JR: Absolutely, ad sales can be made based on im- optimal. Pandora, Spotify, Slacker and many others came TVs, Sonos, Rokus, Logitechs and all sorts of other pressions or on concurrency. For some of our pub- along whose only business is online audio adver- boxes out there, connected dashboards are next. If egta: So you can’t do anything with a user’s previous lishers we actually produce in-market ratings, so tising, and they created the market all by them- you can only measure listening on smartphones, browsing data, using the techniques usually applied if you were looking for a gross rating point target selves. you’re missing a lot of the potential action out to online behavioural advertising, for example? on your schedule, we can provide you with daypart there. And I think we’ll see similar things happening ratings and you can sell it just like traditional radio JR: We don’t use that in our Webcast Metrics, but throughout Europe, where stations have been egta: You use census data for measurement, whereas or television. it is a part of our advertising side, where we do slow to break their simulcast – most of the online radio is typically measured using samples, surveys ad serving and programmatic advertising. We are egta: These sources clearly provide very good insights audio that we see around Europe is pre-roll adver- and panels, and these techniques could also be ap- doing programmatic or behavioural targeting and on listening behaviour, but how can this be matched tising. But we know there are companies that are plied to online audio listening. What are the draw- contextual targeting and retargeting, and all of up with demographic data to understand the audi- beginning to do mid-roll ad insertion as well, and backs of taking a panel approach to online audio? that is basically being done using either cookie ence itself? in some places it’s the only way to do any kind of synch or device ID (being pulled from mobile de- local advertising. I also believe that as companies JR: There are both technical and commercial issues JR: That’s a publisher-by-publisher answer, it de- vices), so it’s not reliant on having to with anything like Spotify turn up their online audio advertising with trying to use samples to measure online au- pends on whether they’re providing us with their with the measurement business. efforts around Europe, marketers and advertis- dio. At a technical level, it’s not a behaviour that’s registration data or not. Generally those that are ing agencies will be eager to access that inventory at the same kind of scale as broadcast listening, That’s a whole different business; it’s really a to- collecting that have gender, year of birth and post-

Page 34 Page 35 tally different method of buying and a different buyers in a very typical broadcast way, based on a worth absolutely zero. Secondly, most of the lis- paradigm. There’s some linkage between the two concurrent listening estimate, and these transac- tening to streaming is happening on mobile de- in our case, because we use the Webcast Metrics tions look a lot like radio. That’s starting to change vices, and ad blocking on mobile devices is far less signal that we get from publishers to help us fore- in the last couple of years. One of the reasons for prevalent than it is on desktop browsers. cast impression delivery on the programmatic that is that we brought a programmatic audio egta: Do you have techniques in place to identify side, but it’s really two different buying or transac- product to market a couple of years ago that en- other non-human activity? tional paradigms: typical broadcast buying on the abled the kind of targeting and accountability that one hand and digital buying on the other. digital media buyers expect, so making audio safe JR: Yes, there are some editing rules that help us for a digital buyer. There’s still some work to be eliminate non-human activities. The first thing we egta: Is it possible to give an estimate of the relative done on post-listen attribution and replacing click- do is apply a filter that removes any activity that size of these two buying markets at the moment? through as the main KPI. comes from IP address ranges on the IAB spiders JR: In the US, and I think this is still true every- and bots list, which is published quarterly. The egta: Again, comparing to online display where ad where in the world, it is still predominantly the second thing, we don’t count any sessions that are fraud is becoming a major issue, do you see the same traditional audio buyers who are now buying less than 60 seconds in duration. Think about the challenges arising in digital audio? streaming audio. Some digital budgets are starting way Google indexes the web, it basically clicks on to move into the space, but the reason why that’s JR: The beauty of digital audio is that there’s vir- links and launches web pages and then moves on been difficult is because audio traditionally hasn’t tually no fraud, in fact it’s hard for me to imagine – it will launch streams just like it clicks on other had all the same capabilities that digital media how you could actually have it. There’s one linear links. But typically we see all of that non-human, buyers require. stream of content, and the stream gets interrupt- robotic activity is very short durations, 35-45 sec- ed by advertising for a period of time, usually – at onds maximum, and by the time you get to 50 sec- egta: For the programmatic side, what tools are you least on the pure play side – not long enough for onds all of that’s out of the system. using to track listeners in order to sell and deliver ad- the listener to care that much. vertising? egta: Can you identify some areas that publishers The tune away from the ads is usually much lower and clients would like to see further developed when JR: Cookies and mobile ad IDs, for example Apple’s on the pure play music services than it is for the it comes to measuring online audio? IDFA or Google’s AdID. IDFA relates to your Apple broadcasters. But that said, it’s not as if there is a account, not your Apple device. We retrieve those JR: I think on the publisher side, the guys who are multitude of ads that are running below the fold, ad IDs from a listener’s mobile device and we ex- larger and have real scale would love to see more as it were, and being reported as viewable; typi- pose those into the exchange and look to see if granular and flexible reporting, so you could cre- cally, audio ads are reported only when they have there’s anyone who’s interested in advertising to ate custom trading zones by client; a particular actually played out – they actually had to be heard that listener. retail client may have three or four regions within by somebody in order to be reported. a market where their stores are located, and they egta: In digital display, the click-through rate is still egta: Do you see the industry suffering any negative might want that kind of flexibility of reporting that often considered to be the king of the KPIs: is this a effects from ad blocking software? doesn’t exist in the space at the moment. From problem for online audio, and are there other perfor- the advertiser/buyers’ side, I think the number mance metrics that would be more appropriate? JR: There are some ad blockers that can be used to one un-met need is for a way to de-duplicate block ads on certain pure play streams, but firstly, JR: Click-through is a tough one; it’s the most com- cume and really provide for a unique reach number when these are in place it’s not a great experience mon KPI in the industry and it’s probably the worst around streaming audio publishers. for the listener, because the software might do its one for audio! Everybody is struggling to find out job but it just makes for silence. It’s not like on a what’s the right performance metric for audio, but web page where if you run an ad blocker you still I don’t think we’re there yet, and that KPI contin- get the content; on a music service you just don’t ues to be the way we’re judged, which is not great hear anything. The pure players may just disable for audio generally. I would say the vast major- those listeners, because an ad blocked listener is ity of transactions are placed by broadcast media

Page 36 Page 37 tising across potentially its entire distribution. Also known (now Nielsen Audio). The PPM is de- signed to be carried Glossary as machine data or return path data (RPD), census mea- by an individual panellist, and it records media exposure References Categories of online audio surement is in contrast to sampling, which collects data both inside and out of the home. 1: Pandora (22 October, 2015). Pandora Reports from only a subset of users and/or devices. The census 3Q15 Financial Results. publishers/platforms: may, however, include all measureable usage from either Reach 2: Mobile Data Report Q2/2015. Personagraph. • Online radio: Radio stations accessible online. These The number of unique listeners to a station or All radio/au- a sample of users or all users. 3: Archer, A. (5 October, 2014). “Introduction to may include live simulcasts of terrestrial stations, dio for a given period of time (e.g. week, day, AQH, etc). The Deterministic and Probabilistic Identification.” brand extensions and linear pure-play Internet broad- Hybrid audience measurement reach of an advertising campaign is the number of unique Coull. casters. In the context of audio, hybrid measurement refers to the people who have come into contact with that advertising 4: Schiff , A. (9 April, 2015). “A Marketer’s Guide To Cross-Device Identity.” AdExchanger. • Online radio aggregators: Services that allow users use of two or more distinct types of audience measure- message at least once. access to live and on-demand audio content from a ment methodology (typically sample-based and census- 5: Edison Research (May 2014). Share of Ear, May 2014. number of different publishers. level) to deliver convergent audience ratings to account for Sample (including panel) based • On-demand music/audio services: Services that al- content that is delivered by both broadcast and over the measurement 6: comScore. comScore Unified Digital Measurement™ Methodology. low users non-linear access to specific audio files, such Internet within a single output. Ultimately, hybrid mea- Representative survey sample from which data is collect- as music or video tracks or podcasts. These services surement may be used to cover both linear (live) and non- ed over time. Sample-based methodologies include CATI, 7: Inside Radio (16 April, 2015). “Countdown to Total Audience Measurement.” may also offer some form of linear streaming with linear (on-demand) consumption. CAWI, CAPI and other questionnaire/interview techniques 8: Triton Digital (2014). Triton Digital AAS/AQH playlists created by experts or through the use of al- and solutions that employ panels, such as diaries and compared to Traditional Radio AQH. gorithms. IDFA electronic measurement (e.g. PPM). Audience measure- Identifier for advertisers, a temporary device identifier 9: Arbitron (2009). Understanding and Using Radio • Personalised radio services: Services that offer linear ment using sample methodologies provides geographic Audience Estimates playlists or customisable radio stations, usually based used by Apple iOS devices. IDFA replaced Apple’s earlier information about listeners and users, but statistical 10: Google. Advertising ID. on an activity, music genre, artist or song title. Prefer- UDID (unique device identifier), which was phased out modelling is required to estimate total usage. ences are typically defined through the use of like/dis- partly due to privacy concerns. Sources for charts, page 6 like (thumbs up/thumbs down) interaction. SDK 1: Edison Research (2014). Share of Ear, May JIC Software development kit, a collection of software devel- 2014. Joint Industry Committee, an independent body repre- Advertising ID opment tools that – in the context of audience measure- 2: Edison Research/Triton Digital (2015). The A user-resettable, unique, anonymous ID for advertising, senting all interested parties, including publishers/broad- ment – allow publishers to integrate the necessary tech- Infinite Dial 2015. provided by Google Play services10. Google Advertising ID casters, media agencies and advertisers, that commis- nology to enable audience measurement organisations to 3: NLO (2015). Audio Distributieonderzoek. replaced the company’s earlier Android Advertising ID. sions and oversees audience measurement. identify and track user access to their audio streams or 4: Spotify (2015). The New Audio. other content delivered online. API PII 5: RAJAR (2015). RAJAR Data Release Q3 2015. Application programming interface, a set of routines, pro- Personally identifiable information, any data that could TSL 6: RAJAR (2015). MIDAS Spring 2015. tocols, and tools for building software and applications. potentially identify a specific individual. Time spent listening, the average amount of time spent Podcast by an average listener for a selected period of time (e.g. Legal Notice AQH week, day, AQH, etc). Average quarter hour, the average number of people lis- An on-demand audio format usually consisting of a digital This publication has been researched and produced tening to a particular audio stream for (typically) at least file (e.g. M4A, MP3, MOV, MP4, M4V, PDF, and EPUB). The UID by egta’s radio department, and it draws inspira- five minutes during a 15-minute period. May also be ex- name is a portmanteau of (i)Pod and broadcast, highlight- User ID, which is used (inter alia) by Facebook to facilitate tion from the team’s conversations with industry pressed as AQH Reach % as a percentage of the population ing the important role of Apple’s technology in its early de- targeted advertising campaigns. experts and literature from multiple sources. being measured and AQH Share % for a given period. velopment and distribution. Podcasts may be downloaded or streamed. Whilst every effort has been made to ensure the Census-level measurement accuracy of the information in publication, egta In the framework of audience measurement, census-level PPM does not accept responsibility for errors or omis- measurement refers to the complete collection of con- Personal People Meter, the trade name for a proprietary sions. sumption data for a particular piece of content or adver- electronic measurement device developed by Arbitron

Page 38 Page 39 Acknowledgements more on this topic egta extends its thanks to the individuals and com- Editorial committee: panies that provided information and expertise for Matthew Carver this publication, as well as their colleagues who Head of Insight supported egta in many ways. E: [email protected] egta would like to thank Triton Digital in particular Yuri Loburets for the company’s support for this report. Director of Radio E: [email protected] This publication has only been made possible through the contributions of:

Arnaud Annebicque (Médiamétrie), Mark Barber (Radiocentre), Jean-Paul Dietsch (OJD), Rob Favre (Triton Digital), Sonya Ford (Nielsen), Kurt Hanson (RAIN News), Henriette Hoffmann (RMS), Jerry Hill (RAJAR), Kristin Koziarski (Triton Digital), Olaf Lassalle (agma), John Rosso (Triton Digital), Julie Terrade (Médiamétrie).

With thanks to:

more on egta

egta 22, Rue des Comédiens, boîte 4 1000 Brussels Belgium T: + 32 2 290 31 31 www.egta.com @egtaconnect

© Copyright egta 2016. All rights reserved

Page 40