The Future of Media Series

Size: px
Start display at page:

Download "The Future of Media Series"

Transcription

1 The Future of Media Series AUDIENCE-BASED TV AD BUYING USING SOCIAL MEDIA DATA vol 5 NOVEMBER 2016

2 Contents INTRODUCTION...3 TV Ad Targeting Today... 4 The Issues with Buying Demos... 6 The Advantages of Audience-Based Buying... 8 Leveraging Social Media for TV Ad Targeting... 9 The Future of TV Advertising...11

3 Introduction It s clear that the way TV advertising is bought and sold is evolving. TV is learning how to implement some key elements of the digital advertising playbook by making its processes more efficient through new data-driven and automated techniques. In the mid-2000s, digital advertising buyers and sellers began transacting via programmatic auction marketplaces to meet needs on both sides. For online publishers the programmatic revolution helped monetize remnant inventory that was unsold due to the inefficient processes of selling ads manually. On the buy side, programmatic impressions available at scale enabled marketers and their agencies to bypass high margins from ad networks and access inventory directly to apply their own data targeting to achieve better ROI. This revolution in the online ad market led executives to wonder about the applicability of such methods to the TV ad market. However, there are a number of very important differences between the TV and digital ad markets. Perhaps the main one is that although TV advertising market practices and workflows may not be showroom fresh they certainly aren t broken. As we described in our first white paper, TV advertising was built upon processes designed as far back as the 1950s, but it still works fairly effectively for both buyers and sellers. The 2016 upfronts provided evidence of the strengthening US market with networks boasting CPM gains, however it s clear that the trends around multi-screen consumption and delivery are top of mind. We believe that the key driver in the evolution of TV ad buying and selling is the desire of brand marketers to use data-driven technologies to improve targeting beyond the legacy standards of today s demographic buying techniques. As discussed in our last white paper, Programmatic TV is a long way from reaching scale but there are opportunities that can be activated today. Volume 5 in our Future of Media series takes a closer look at new developments in advertising targeting and how these can deliver immediate benefits. 3

4 TV Ad Targeting Today By and large, when a brand procures a TV ad they re trying to make people aware of their product or service and persuade them to buy it now or in the future. Imagine you re about to launch a new candy bar or car and you re in charge of the ad budget. Traditionally you would have sat down with the media planners at your ad agency and designed a TV campaign comprised of ads, often bought way in advance, based on a number of factors that hopefully provide a best-possible guess for an audience that is likely to want to buy your new candy bar or car. This decision-making process would likely be based upon these attributes: Demographic viewing data Geographical area Research and assumptions about likely habits and interests of a particular program s viewers Research and assumptions about likely habits and interests of viewers during particular dayparts Unfortunately, many of your ad views would be wasted, being watched by people without a sweet tooth, by non-drivers, or by drivers that have just bought a new car. 4

5 Clearly, there are issues with the current method of targeting. For one, the demographics are hardly granular, splitting everyone into two genders and a handful of age groups (e.g. men aged 18 49). In mature markets, including the US and UK, viewing data is collected by digital people meters which automate viewing data capture, although there are issues we will see later over the relatively small sample sizes. In some territories data is still primarily collected using unreliable and inaccurate paper diary-based methods, occasionally augmented by smaller panels of more accurate frequency-based meter-collected data. Meanwhile, daypart data and geographical buying may be useful but are still broad proxies for the audience you really want candy bar buyers that are likely to be attracted by your brand values or people looking to buy a new car in the relatively near future. However, it turns out that accuracy of data was never that important. The way that TV ads were traditionally bought and delivered meant that highly-granular audience data were just providing a level of detail that audience buying could not match. It s a little like a golfer aiming where to hit the ball: if you only land on the green 50% of the time from 100 yards there s hardly any point aiming at a particular 3 or 4 square inches with this level of accuracy you re much better off aiming for the middle of the green! In this white paper we ll look at how new data-driven techniques can deliver far better results now, even with today s relatively un-targeted ad delivery methodologies. We will also look further into the future at how data will be part the inexorable (but not necessarily fast) journey towards targeted and automated ad planning, buying and delivery. 5

6 The Issues with Buying Demos To understand what people are actually watching, the most common method is to monitor the TV viewing habits of a sample of households. In general, an invitation to become a panelist is generated with the aim of broadly matching the demographics of the nation. These small subsets of panelists are then scaled to cover the entire country. FLAWED PANELS PRONE TO ERROR The 22,000 strong panel used by Nielsen in the US represents 116 million American TV families. In the UK, the 5,000 households used by BARB are intended to represent the habits of 26 million TV households. Each person in the household that enters the TV room signs in through the remote control and the frequency-based meter then records who is watching what. This data is collated by the audience measurement agencies to provide overnights the previous day s TV viewing figures per show. Some of these measurement systems also include those who watched the show at the time it was broadcast and also those who recorded it and watched it back the same day. Participants in these representative panels are often asked to fill in additional surveys that ask questions about lifestyle choices, hobbies and attitudes towards things like airline travel, fitness and holidays. These panel approaches, as exemplified by Nielsen, were developed in the 1950s when the recording of viewer data was collected on paper-based viewing diaries. This technique has been developed to include meter-driven data capture and to record new viewing opportunities such as pay-tv, satellite and viewing over non-traditional TV devices such as smartphones, tablets and laptops. However, there are issues: the sample-sizes are small, they are not necessarily representative of the whole population and the measurement of web and mobile is in its infancy. 6

7 A FEW VOICES SHOUTING LOUDLY In the UK, the sample-sizes are such that a panelist is proxy for 5,000 households or more and it s not that much better in the US. Yet thinking about the different types of household types using just three criteria of nationality (196 countries), ethnicity (the UK 2011 census lists 18), and gender gives you over 7,000 combinations. This means that in a country like the UK, the TV viewing habits of say the 700,000 women of Indian decent may be interpreted through the actions of just 10 panelists. From the sell side, a niche channel with 150,000 viewers may again be based on the sampling of 10 people or less, and for the really niche channels viewing figures can fall off a cliff when a single panelist that loves your channel goes on holiday. A SAMPLE OF PEOPLE WHO WANT TO BE SAMPLED The fact that panelists have to opt-in provides further areas of concern. Although selection criteria can vary, participation is entirely voluntary. This leads to concerns that many groups, such as ethnic minorities, younger and much older audiences, undocumented citizens, renters, low-income workers, and those with lower educational attainment, may be undermeasured by the panel approach. TODAY S WORLD IS MULTI-MODAL There are a lot of screens that people are watching TV content on that are simply not being measured. All panels vary in the ability and accuracy of capturing data on who is watching what on which screen, but the consensus is that none seem to perform particularly well. To be sure, panel based measurement should not be written off. It provides a recognized and well understood currency for the TV advertising industry, even if that currency is increasingly under attack from those wanting more accuracy. When added to survey data and third-party data, panels provides a broad baseline of demographics. That said, many believe that the combination of the proliferation of channels and new ways to watch TV diminishes the long-term accuracy and relevancy of the panel approach. 7

8 The Advantages of Audience-Based Buying The digital ad market has taught us the value in serving ads based on a good understanding of the person watching a screen. Before you can move from planning a TV ad campaign based on demographics to buying a TV ad audience based on your particular criteria, you need to build an understanding of the consumers correlation with the TV shows that they watch. To do this requires data, but it s important to remember that not all data sources are created equal. Some of the legacy data sets that have historically fueled TV targeting such as sales and CRM data are largely representative of existing customers and not necessarily representative of the full market. This has potential for overspending to reach existing customers. Further, consumer survey data, another traditional data set for TV targeting has a set of problems with small survey sizes and nonrepresented groups leading to the same kind of biased misreporting that occurs in TV panels. Even third-party data sets may have limited authority as the provenance is difficult to assure. So what data is available to provide the precision and scale needed to enable advanced TV buying? 8

9 Leveraging Social Media for TV Ad Targeting Social media can provide the precision and scale needed for advanced TV buying. 4C founder and chief scientist Dr. Alok Choudhary refers to this as the limitless focus group. Social data can enable TV advertisers to target customizable audience segments based on real life actions, interests, and connections. Social datasets are rich in observed behavior on what consumers are saying, doing and sharing. With a pool of two billion+ social media consumers, brands and advertising agencies can power audience modeling tools at scale and in near real-time. HOW IT S DONE The following real-life example shows the power that social media data can offer and how they enable audience buying for TV advertising today. Turner came to 4C to see how we could use social media insights for an advertiser that was promoting a movie launch. The client knew that social buzz is a key factor in driving box office results and previous 4C research with Turner showed that TV advertising is the second largest driver for social media engagement. To increase awareness for the release of the major motion picture, we partnered with Turner to deliver the first industry solution for targeting TV audience segments utilizing social media data. 9

10 STEP 1 The approach we designed was was modeled off similar movies knowing that would be as a fundamentally better proxy for likely interest than using demographic ad buying. Our data science team first assessed movies that were similar to the new one or that would be very likely to appeal to the same audience as the new movie. STEP 2 By analyzing social media engagements from recent successful films that were similar to the new film, we identified people who showed a pattern of engaging with those films. Through mapping the social behavior of those individuals to TV viewership at the respondent level we created a targetable audience segment for TV advertising, or what we call a TV Fingerprint. STEP 3 At this point the TV Fingerprint can be used to buy the campaign through traditional TV ad buying workflows. In this case, Turner used its own TargetingNOW solution to plan the optimal TV schedule for a fourweek marketing campaign leading up to the movie release. To be sure, the ad campaign could just as easily have been planned through Nielsen Strategic Households or any other commonly used TV ad planning platform using the 4C TV Fingerprint. RESULTS The results were measured by analyzing the lift in social engagement that occurred during the period two minutes before and two minutes after each TV commercial for the film aired. This data was collected across each TV ad that ran for the entire film campaign on all TV networks. We then aggregated the lift scores for each TV ad and calculated Turner s average lift versus other cable networks average lift. This groundbreaking method of using social media data to target TV audiences drove nearly 3X more social media engagement for the film across the optimized Turner schedule in comparison with other cable advertising during the same period. 10

11 The Future of TV Advertising BLUNT TOOL OR SHARP SCALPEL? Accurately predicting the future is never easy. But we don t think we re sticking our necks too far out in saying that TV ad buying is heading in the following directions: Increased targeting, primarily delivered by datadriven techniques Increased automation Increased addressability More accurate and quicker measurement Better integration in terms of planning, buying and delivery with other advertising mediums. With this in mind, marketers should move away from solely demographic buying of TV ads. You don t have to wait for infrastructure to be built out. You don t need broadcasters to invest in new ad delivery technologies. And you certainly don t need to wait for the rollout of full programmatic TV advertising; exciting though those solutions will be. The time is now to start replacing the rather blunt tool of demographic buying by the sharp scalpel of social media data-driven audience buying. Our methodology has proven that it can deliver great results and things will only get better as targeting, automation and addressability permeate the entire industry. As we said in our second whitepaper in this Future of Media series, TV + Social is a 1+1=3 equation. In the case of our campaign with Turner, it equaled 3x. What brand doesn t like that math? 11

12 About 4C 4C is a global leader in data science and media technology with solutions for multi-screen convergence. Brands, agencies, and media owners rely on 4C to improve effectiveness across channels and devices. The 4C product suite includes activation on Facebook, Twitter, LinkedIn, Pinterest, Instagram and Snapchat as well as TV Synced Ads across display, search, social and video. The company also provides advertising and content analytics leveraging its Teletrax global TV monitoring network and proprietary social affinity database. Founded in 2011 and based in Chicago, 4C has staff in 14 worldwide locations across the United States, United Kingdom, the Netherlands, France, Hong Kong, India and Singapore. Visit for more information.