Advertising: The “Golden Age” of TV Enters its Golden Years
SEE LAST PAGE OF THIS REPORT Paul Sagawa / Artur Pylak
FOR IMPORTANT DISCLOSURES 203.901.1633 /.1634
psagawa@ / email@example.com
February 3, 2015
Advertising: The “Golden Age” of TV Enters its Golden Years
US TV advertising has crested and is beginning its inevitable decline. In an era of rising transmission fees and a vigorous market for content licensing, media companies have been reporting disappointing revenues for their TV network units – the result of poor ad sales. The networks drove through rate increases during last May’s Upfronts at the cost of reduced volume, a strategy that appears to be backfiring after poor ratings during the fall season and a correspondingly weak scatter market. Meanwhile, Nielsen, whose methodology we believe is significantly biased toward over counting, acknowledges declining viewership for channelized TV, while advertisers decry the accompanying deterioration in the attentiveness of that audience toward their commercials. These viewers are migrating to streaming video, in their living rooms as well as on their mobile devices, and advertisers are shifting their attention to digital as well. The extraordinary trajectory of online advertising, reflected in the double and triple digit ongoing growth in ad revenues at GOOG, FB, and TWTR, has begun to eat into TV’s piece of the pie. Big advertisers, in traditional categories like autos, consumer products, telecommunications and financial services, are explicitly stating their intention to shift budget dollars from TV to digital, while broader surveys of marketers suggest the same thing.
- US TV ad spending has been decelerating. 2014 US TV ad sales were up 3-4%, in an Olympic and election year, after 2-3% growth in 2013. This is a deceleration from 6-8% growth in the 2012 election year and 4-5% in 2011. Ominously, 4Q14 sales were down ~2% YoY, as weak Upfront sales drove dependence on the scatter market during the fall season, exacerbating the deterioration in TV ad pricing, which was off 1% for the year and down more than 12% since 2009. The falling rates have been offset by increasing the number of ads shown, up 6-8% in the same 5 years, and by a growing number of TV households. The increasing advertising load has the unfortunate side effect of potentially alienating viewers who have other options, and of raising the likelihood that ads will be ignored or skipped.
- Active TV viewing is clearly in decline. In 3Q14, Nielsen finally acknowledged the obvious – US per capita TV viewing is in decline. Still, we believe Nielsen’s methodology seriously overstates both the size and quality of the TV audience. Even with the reported downturn, Nielsen still estimates that Americans actively watch over 5 hours of TV every day – an increase of 12 minutes since 2010. Meanwhile, streaming video has more than doubled to nearly 45 minutes a day over the same timeframe, a trajectory corroborated by multiple sources and inconsistent with Nielsen’s story of a growing TV audience. Moreover, it has grown increasingly likely that users assumed to be watching, because the TV is on, are not actually paying attention. The quality of attention is particularly suspect for commercials – DVR users – nearly a third of the prime time audience – routinely fast forward through ads, while live viewers use the increasingly ample breaks to peruse alternative media online.
- Broadcast TV audiences continue to decline. Nielsen’s Broadcast Primetime C3 metric (inc. 3 days of DVR/on-demand viewing) for the fall season was down -9% YoY and, over the last decade, the C3 ratings have fallen more than 41%. This has seriously eroded the major selling point for broadcast TV as the only medium able to reach a 10%+ swath of Americans at the same time. 10 years ago, 8 different weekly series averaged better than 20M viewers a week. Today, only sports and one-time events can hit that mark. Broadcasters complain that growing use of DVRs makes the C7 metric, which would increase primetime ratings by ~3%, more appropriate, but buyers, given that 35% of ads are time sensitive and the propensity for time shifted viewers to skip ads, have resisted. Only GroupM, representing a pool of advertisers, has agreed to C7 and then, in exchange for a sharp discount on pricing.
- Streaming is already >12% of US video viewing and growing prodigiously. NFLX delivers nearly 2.1B streaming hours/month in the US. Assuming each stream is watched by an average of 1.5 people, spread over the 262M Americans living in HDTV equipped households, NFLX is alone accounting for 12 hours of monthly viewing per person. YouTube, which streams more than 1.5B hrs/mo in the US, likely has fewer viewers per stream, but still counts for at least another 6 hrs/mo per person. Add in Amazon, Hulu and other smaller streamers, and total US online video viewing likely exceeds 22 hrs/mo, tops 12% of total video consumption, and is growing at a better than 30% annual rate. With big investments in high quality original programming and with access to online content getting much easier, the pressure on channelized TV will get worse and the temptation for advertisers to shift their budgets toward the Internet will grow. It will also raise the stakes in negotiating for new content, raising costs for networks even as online content shrinks their audience.
- Advertisers shifting budget to digital. SMI reported 4Q14 digital ad spending up 15% YoY, compared to a -2% hit to the TV spend. Online media have introduced richer ad formats and more sophisticated targeting and tracking mechanisms, while audience measurement and ad tech companies have delivered improving tools to evaluate effectiveness. Meanwhile, moves by MSOs and TV networks to introduce new technologies to blunt the online advantage, like digital ad insertion, have been painfully slow. We believe that the inevitable future growth in online ad spending will come directly from TV ad budgets, portending a slow but inexorable slide in revenues for network ad sales.
- The May Upfronts could be a significant negative catalyst for media stocks. While networks declared victory at the 2014 Upfronts on CPM increases, sharply lower volumes left them vulnerable to the scatter market, and thus, 4Q14 revenue declines. We expect this year’s Upfronts to be demonstrably weaker. Advertisers may draw a line in the sand vs. the network push toward C7 ratings and their lower quality audiences, yielding both lower pricing and volume commitments. With this, we see the risk to reward ratio for advertising driven media stocks – e.g. SNI, DISCA, CBS, FOX, VIA, etc. – as distinctly unfavorable. We continue to prefer digital advertising, with GOOG, FB and TWTR viewed as structurally advantaged.
Fear the Upfronts
Traditional TV hit a crossroads in 2H14. Industry lapdog Nielsen finally called out a YoY decline in viewership after more than a decade of sliding primetime broadcast ratings and years of rapid growth in streaming video viewership. Metrics also show that more than 30% of viewing for popular programs is time-shifted (via DVR, on demand, or streaming on network owned sites), greatly reducing the quality of the audience for advertisers. Not surprisingly, The Standard Media Index’s estimate of US TV ad spending dropped 2% YoY in 4Q14, suggesting the first annual decline since the 2008 recession despite campaign spending and an increasingly strong economy. Weak volume sales at last year’s Upfronts have both broadcast and cable networks at the mercy of a soft scatter market, where advertisers wait until the last minute to scoop up bargain inventory. This situation will not get better.
The ad community is growing increasingly enamored with digital formats, where specific viewers can be accurately targeted and their behavior tracked from before engagement through to purchase and beyond. As viewership of online streaming video continues to blossom, and as popular social media like Facebook and Twitter expand their video offerings, competition for ad spending will only intensify. While networks are shifting to accommodate this by selling programming to online services and by launching their own “over the top” products, the pain of losing dominance in distributing video content will be far more acute than the benefits to be gained in expanding their role in the digital market. Networks hope to salve that pain by pushing to include 7 days of deferred viewing in the audience calculations for ad fees (C7 ratings), but advertisers, wary of ad skipping and the loss of timeliness, have largely resisted.
This year’s Upfronts could be ugly. The argument for “TV Exceptionalism”, asserting that TV advertising is unique in its ability to deliver a rich media message to large, coveted consumer cohorts within a short timeframe, has been punctured. We expect push back on CPM demands, hesitancy on volume commitments, and refusal to adopt C7 ratings without considerable compromises on price. This will not be good news for ad supported television, either broadcast or cable. Odd years, without national elections or Olympics, are traditionally weak years for TV advertising, and this year, given sea change shifts in the landscape and despite the recovering economy, we are likely to see TV ad sales down.
Of course, those closest to the matter may be the last to see it. Amongst advertising industry analysts, Magna Global is alone in calling for 2015 US TV ad sales to fall. Ad driven media executives are talking a brave game, and their relentlessly upbeat Upfront presentations will undoubtedly tout a rainbow behind the cloudy sky. Still, after years of relatively strong stock performance, it may be time for the reality to finally sink in. Weak TV ad sales in 2015 will not be an aberration, but rather, the start to a long and inexorable decline. Stocks that are particularly exposed to that slide – e.g. SNI, DISCA, CBS, FOX, VIA, etc. – will find it difficult to make up the difference with their own digital efforts or through higher fees from distributers facing their own digital threats. We continue to prefer digital advertising, with GOOG, FB and TWTR viewed as structurally advantaged.
Exh 1: US Ad Spend by Major Media Type, 2011 – 2014
Exh 2: Summary of 2015 US Advertising Forecasts by Media
Exh 3: Impact of shift from C3 to C7 ratings on select major programs
Exh 4: Primetime TV Season Ratings, 2013-14 versus 2012-13
Asking More for Less
“Don’t get too attached to any of these shows, because most of them won’t survive. It’s like adopting a kitten with cancer.”
Jimmy Kimmel hosting ABC’s 2014 Upfronts
TV network owners negotiate deals for most of their commercial time in advance, during “upfront” negotiations that usually take place in May and June following star-studded presentations to the advertising community showing off their fall season lineups. Network executives typically look for a CPM (cost per thousand impressions) increase, in hopes of driving ad revenue growth. Media companies have also been long used to strong steady growth in TV CPMs, acquiescing to network CPM increases in the mid to high single digits over the past 5 seasons (Exhibit 5). Even as many media categories – newspapers, radio and yellow pages in particular – have suffered badly from Internet competition since the turn of the millennium, TVs share of the total ad pie has only increased. Excepting the understandable recession driven 22% drop for the 2009 season, TV ad receipts have increased every single year since the previous recession. However, the growth trajectory has shown unmistakable signs of slowing with 2-3% ad revenue growth for CY13 the lowest non-recession number in decades, followed by relatively anemic 3-4% growth in the ordinarily strong Olympic/election year market in 2014. This is part of a pattern of deceleration that has been playing out since 2010. This bodes poorly for CY 2015. Traditional advertising forecasters ZenithOptimedia and MagnaGlobal are split in outlook of the US TV ad market this year, forecasting an increase of 1.7% and decrease of -1.4% respectively. We think projections of a rising TV spend are wishful thinking.
Exh 5: Broadcast Upfront Results, 2007-2015 Seasons
Results from the last two TV Upfronts indicate a turning point in television advertising, with aggregate upfront commitments falling -8.3% since the 2012-13 season. The Standard Media Index (SMI), which reports data directly from agency booking systems on a monthly basis, has seen the level of upfront revenue has fallen precipitously. All advertising not placed during the upfronts, also referred to as “scatter,” now makes up 23% of revenue for cable networks and 16% of broadcast ad revenues. Scatter’s revenue share has grown 600bps and 500bps respectively over last year (Exhibit 6-7).
Exh 6: Scatter Market Revenue as a % of Total
Exh 7: Cable Network ‘Upfront’ Advertising Commitments
Stopping the pattern is an uphill battle. Primetime ratings for the broadcast networks, as reported by Nielsen, have been falling for decades, ravaged first by the rise of cable networks and more recently, by online streaming video. The broadcast primetime C3 rating metric, which measures live viewing and time-shifted viewing of programs on the 6 broadcast networks during the three day window following the original airing, was off 9% YoY for 4Q13, and down 41% vs. the prior decade. Because of this slide, the average cost of a 30 second spot in primetime is down 12% since 2009 despite the steady increases in the CPM (Exhibit 8). Networks have responded to this by adding additional commercial time to each hour of content, with the broadcast nets raising their ad load from 13 minutes and 25 seconds per hour in 2009 to 14:15 today. With almost 25% of every hour now devoted to ads, many viewers may be finding their tolerance tested.
Exh 8: Commercial Clutter and Average cost of a 30 second spot
Exh 9: Average Network Prime Time Households and CPMs, 1980-2014
With audiences in inexorable decline, and the advertising load near a practical threshold, the networks are grasping at straws (Exhibit 9). One such straw is a proposal to change to the C7 ratings metric, which would increase the window for DVR and on-demand viewing out to a week. Given an ongoing move toward time shifted viewing, such a change would add roughly 3% to overall primetime ratings and increase ratings for some popular dramas by as much as 10% (Exhibit 10). However, most advertisers are pushing back hard. An estimated 35% of TV advertising is considered time sensitive – think opening week movie ads or automobile sales promotions – and impressions a week later for these spots are considered nearly worthless. Furthermore, more than half of DVR viewers skip through the commercials, while the remainder may be failing to do so because of distractions. GroupM, the powerful ad buying arm of the huge WPP agency, has reportedly signed deals for C7, but at significant CPM discounts.
Exh 10: Financial impact of extending ratings to C7
If the networks can engineer an industry wide transition to the new metric, it will come at the cost of much lower CPMs for time shifted impressions, making the industry revenue impact well below the estimated 3% bump in total viewership. Moreover, it is almost certainly a one-time bump – it’s hard to imagine advertisers agreeing to a C10 measure for primary ad buys.
Hence the air of puffery and bravado. If audiences are in inexorable decline, the advertising load is near a practical threshold and a shift to C7 will add little IF it can be accomplished, CPM is the only lever, and both networks and advertisers know it. Given ad skipping on DVRs, mobile device multitasking during commercial breaks, and the inability of Nielsen’s metrics to distinguish between active from ambient viewing, buyers have ample reason to question the continued efficacy of TV advertising.
When is a Viewer Not a Viewer?
Nielsen’s rating methodology relies on devices called “peoplemeters” that connect to each TV set in a population of roughly 50,000 US households, selected for ethnic and geographic diversity. Each “peoplemeter” comes with a remote that controls the choice of channels on its TV and has a separate button for each member of that household to log in as they begin and end watching that TV. The sample households are recruited with an offer of quarterly payments (~$50-100) and Nielsen supports the household with regular visits to assure that the system is operating as intended. Some peoplemeter families have reported developing friendly relationships with their Nielsen representatives, who often bring food or other token gifts during their visits. The peoplemeters collect the reported viewing from the household TVs and then upload them each night to Nielsen’s data center, where the data is tabulated and the overnight ratings calculated.
Exh 11: TV Households, 1980-2014
There are several important problems with this methodology, which has been in place for 25 years. The first is sample bias. Specifically, the 50,000 families recruited to have the intrusive peoplemeters installed in their homes, may not be truly representative of the 116M TV households in the US (Exhibit 11). Nielsen keeps the identities of these families and the exact demographics of its sample a proprietary secret, but public filings related to a lawsuit filed by Fox a few years ago claiming that Nielsen’s sample was biased against minority families, it was revealed that the sample actually modestly overweights both African-American and Latino families relative to the overall population. More transient viewers, such as college students and other short term renters, are not recruited, and viewers that watch in a group setting, such as a bar or a common room are not counted. Given that each sample household must agree to a somewhat intrusive process of logging in via the specialized remote, a further bias may be toward families that have unusually strong desires to influence the ratings process, either because of their unusual tastes or higher than usual consumption of TV. Given that families, once recruited, tend to remain participants for multiple years, any underlying sample bias is perpetuated.
The second form of bias comes from compliance. While families are asked to log in each person that is watching the TV at the moment, and to log out each person as they stop watching, particularly if other family members continue to watch, it is difficult to believe that these rules are followed religiously. Thus, a child who begins watching the TV at 8PM and leaves to go to bed at 9PM while her parents continue to watch may remain logged in until 11PM when the TV is turned off. More insidiously, family members may be logged in while in the room, but engaged in an entirely different activity – perhaps even watching video on a mobile device – or they may actually be logged on to two different TVs within the house at the same time while the kitchen TV blares its programming to an empty room.
Finally, the Nielsen survey may be biased because of the relationship between the families and Nielsen itself. By providing regular compensation and establishing regular personal contact between Nielsen staff and the families, the behavior of the families may be influenced. A family may purposely overstate their watching (false log ins, delayed log outs, etc.), or shift their viewing toward programming considered “high brow” (a phenomenon that was apparently rife during the self reported diary methodology that preceded peoplemeters), in an effort to curry favor with Nielsen or to “prove” their worthiness as a Nielsen family.
Despite the opacity of Nielsen’s data and the inadequacy of its methodology, its ratings are unchallenged as the measure of TV audiences and thus, as the measure of advertising reach for the medium. In the emerging cloud era, where internet video purveyors can target advertising at specific individuals, determine whether or not each individual actually watches the ad, and then track each individual’s online behavior after the ad impression, the imprecision and bias of the Nielsen ratings are potential millstone around the necks of TV networks looking to drive the premium prices paid for their ads even higher.
Watch Around the Clock
The propensity of Nielsen’s methodology to over count TV viewership is apparent in its aggregated TV audience estimates. Despite reversing course and showing data demonstrating the decline of TV viewership, Nielsen’s numbers likely overstate media consumption to a significant extent. According its most recent report, in 3Q14, the average American watched more than 155 hours each month, or 5:02 per day, of live and time shifted TV, up from 148 hours in 3Q 2009. Taken at face value, this means that on average EVERY American living in an HDTV capable household, spent more than a third of their waking hours actively watching television, every day of every month of the quarter. Factoring in work and school hours raises the number to more than HALF of all hours not spent sleeping or working are spent in front of the TV. Consider also that everyone has non-working obligations and/or activities that preclude active TV watching, suggesting that the propensity to watch TV, according to Nielsen, may be as high as two-thirds of available time. Add in that this number is a median, and that half of the population has to be watching more than 170 hours of TV every single month. Factor in that Nielsen also estimates that the average American is spending more than 60 hours a month, or another 2 hours a day, on the Internet via their mobile devices or fixed computers, and it’s shocking that people have enough time to shop or eat (Exhibit 12).
Exh 12: Nielsen numbers in perspective, Q3 2014
We are extremely skeptical of Nielsen’s estimates. Its most recent report suggests that Americans spend just 10 hours each month watching streaming video, with most of that attributed to desktop computer usage. However, using other sources reveals a very different story. Netflix recently reported that it streamed 6 ½ billion hours of video during 1Q14, or about 2.2B hours per month. If usage growth is proportional to and grows at the same rate as subscriber growth, we believe Netflix could have streamed 7.7B hours of video in 4Q14. 68% of Netflix subscriptions are domestic, and assuming that Americans, with a higher penetration of connected TVs and with a longer average subscription tenure, tend to use the service more often, we estimate that roughly 2.0B hrs/mo are domestic (Exhibit 13). Note that Netflix is reporting streams but that Nielsen is estimating viewers, so multiplying those 2.0B streaming hours by an assumed 1 ½ viewers per stream, and we estimate Netflix US monthly viewership at just about 3B hours. Dividing this usage by Nielsen’s denominator, 262M US viewers in HDTV households, yields over 11.3 hours of Netflix viewing per person, up nearly 18% YoY, 78% of which the company reports as streaming to a TV rather than a computer or mobile device. Clearly Nielsen families are skewed away from Netflix viewers.
Exh 13: Netflix Streaming Metrics, Q4 2014
Exh 14: SSR Calculated Streaming per US Viewer, Q4 2014
Exh 15: US Streaming Traffic, March 2013 versus March 2014
Exh 16: US Streaming Traffic, 1H13 – 2H14
Applying the same process to YouTube – which reports 6 billion hours of streaming per month, 20% of which are domestic – and assuming just 1.25 viewers per stream, yields 1.5B viewing hours per month. Again dividing by Nielsen’s 262M denominator, we estimate 6.4 hours of YouTube viewing per person per month, up 20% YoY, bringing the per person monthly streaming total to 17.4 hours for just Netflix and YouTube (Exhibit 14). Viewer statistics for smaller streamers are harder to come by, but services like Qwilt and Sandvine provide measures of downstream internet traffic coming from Netflix, YouTube, Amazon Prime Video and Hulu. Using Qwilt’s data, which is more complete, Netflix is responsible for 57.5% of all North American streaming traffic, YouTube 16.9%, Amazon 3.0%, Hulu 2.8% and Xbox 0.5% (Exhibit 15-16). (Note that YouTube’s relative share of traffic vs. Netflix is less than its share of viewership, almost certainly due to a higher mix of less than HDTV streams to mobile devices.) Assuming that their relative share of viewing time is the same as their relative share of streaming traffic, we estimate 3 hours per month per person for Amazon, Hulu and Xbox.
There are a number of other streaming sources – Yahoo, TV.com, AOL, Daily Motion, all of the individual TV network sites (e.g. Fox.com, NBC.com, HBOtoGo, ESPN3, and many others). Assuming that these sources add another 1.1 hour per month per person gives us an estimated total of 20.5 hours for each of those 262M people in HDTV equipped households – considerably higher than Nielsen’s estimate of less than 14. We also note that US streaming viewership has been growing very rapidly, up more than 40% over the past year and up more than 90% vs. 2009. This casts Nielsen’s assertion that US TV watching is up by more than 6 hours per month in the past 5 years in a dubious light. Considering strong evidence that streaming viewing has increased by more than 15 hours per month, the implication is that Americans have increased their video viewing hours by 12.5% since 2009, adding an additional 42 minutes of watching every single day.
Exh 17: Nielsen numbers in perspective by aggregate monthly hours, Q3 2014
We do not believe this is credible. Considering the concurrent and indisputable growth of streaming and social media online, it seems beyond improbable that the average American could still be actively watching more than 5 hours of television every single day of the year (Exhibit 17). More likely, Nielsen is basing its estimates on a sample of households that is highly skewed toward heavy television watching, and that the methodology is counting significant amounts of time for household members that are not actually watching the TV. It is also likely that these biases have grown more acute over time as the online alternatives to TV watching have grown more attractive and available. Moreover, the advertising community is growing more aware of these biases and the implications for the size and quality of audiences that TV can actually deliver.
Despite the likelihood that overall channelized TV watching has begun to wane, the near certain deterioration in the quality of attention paid to TV commercials, and the incontestable decline in primetime broadcast ratings, TV executives show little sign of worry. Their first argument is that TV is just different, that the ability to show a moving advertisement that can be seen on millions of living room screens at the same time is uniquely valuable, particularly for high value, time sensitive messages. Through TV, brands can be tied to iconic events and reach a large cohort of attractive potential customers at a specific time with a single buy. The second argument is that TV has a long track record of success for advertisers and well established, widely accepted metrics (like Nielsen ratings) to verify its reach. Finally, the negative trends are not new and in their face, television networks have grown their ad revenues and as an industry, taken an ever larger share of the overall measured media advertising pie. Media companies and brokerage analysts have begun to view the threat of over-the-top video as akin to Chicken Little’s proverbial falling sky. Deals will get done, just as they always have.
Well, maybe not. Channelized TV is not so different anymore. Netflix reports that 78% of its streams now go to television sets rather than computers or mobile devices, and while Netflix does not run commercials, it may well be a gateway drug to services, like YouTube or Hulu, that do. Much of this has to do with the proliferation of adjunct boxes, like gaming consoles (Xbox, PS4) and alternative set-tops (AppleTV, Roku, ChromeCast), that make it easy to mix streaming video with channelized TV. Furthermore, TV’s ability to reach the masses is compromised by the falling primetime ratings. In the 2003-2004 season, 7 different programs delivered average audiences of more than 20 million viewers, week after week (Exhibit 18). A decade later, only Sunday Night Football, with a scant 18 week run, managed to hit that standard, and no other program would have managed to hit the top 10 ranked shows of ’03-’04 despite there being 8 million additional TV households in the population. This problem is exacerbated by the growing propensity toward time shifting. DVR playback is now more than 25% of the viewing audience for primetime cable programming, and 48% of total primetime broadcast viewing (Exhibit 19-20). Total time shifted viewing has grown nearly 40% over the past 3 years, going from 6.3% to 8.6% of total channelized TV viewing during that time. 50% of those viewers fast forward past the commercials. Little wonder that TV execs are pushing so hard for C7 ratings. Still, from an advertiser’s perspective, particularly for the movie companies, car companies, and other buyers whose messages have a short shelf life, a time shifted viewer is NOT the same as a live viewer.
Exh 18: Shows with average viewership over 20M viewers, by season, 1980-2014
Exh 19: Percent Live versus Playback Minutes, P18-49
Exh 20: Time Shifting in Primetime, P18-49
Certainly, TV has the distinct advantages of incumbency, when it comes to advertising. Agency buyers have relationships with their network counterparts. Clients are accustomed to spending big on TV, and can see their spots plain as day as they run. Nielsen’s ratings, flawed as they are, have been running the same way for 25 years, and agencies have developed their own proprietary tools for tracking and measuring the impact of TV ads. Old habits die hard. Still, the advantages of online media are profound. TV networks offer ad buyers an audience of a predicted size and mix of demographics (age, gender, ethnicity, location, etc.) and if the actual audience varies from the prediction in any meaningful way, the network promises to make good by running free ads to make up the difference. Online video ads can be targeted to specific individual users with precise demographic and behavioral attributes – e.g. consumers who have visited a company’s website recently or who have recently searched for information on new cars. Most streaming video ads allow viewers to “opt out” and advertisers only pay for the ads that are actually watched by consumers. Finally, online advertisers have the ability to follow consumers AFTER the commercial impression to see the impact on purchasing behavior. Do viewers, exhorted to visit a website to get more information, actually go to the website or even buy? Online advertisers know and TV advertisers do not.
Exh 21: Annual Newspaper Advertising Spend versus Circulation, 1950-2012
The final argument – that primetime audiences have been dropping for years, while TV ad CPMs keep going up – is dangerous. The history of newspaper advertising is a sobering example of why this is true. US newspaper circulation began dropping in 1992, yet newspaper advertising receipts continued to rise (Exhibit 21). Newspaper ad revenues took a big hit with the internet bubble recession of 2001, but resumed growth in 2003 and reached a new peak in 2006. Industry executives and analysts during the post recession recovery made their cases for “Newspaper Exceptionalism”. Newspapers were unique vehicles for advertising, went the argument, and, as such, were evergreen despite the deteriorating audience. Local, time sensitive advertisers – movie theaters, auto dealers, grocery stores, classified advertisers, etc. – had no other medium appropriate to their needs, went the spiel. History has been unkind to these arguments. Newspaper advertising sales fell 60% between 2006 and 2009. Many papers have ceased publishing and all of the others are struggling. With this context, it is difficult to imagine that TV can remain exceptional forever.
The Barbarians are at the Gate
The day of reckoning may be upon us. The IAB estimates 1H14 US internet advertising spending at $23.1B, up 15% YoY, while the video component of that was over $3.5B, growing at a whopping 55% CAGR (Exhibit 22-23) with an annual run rate of better than $7B. In contrast, the IABs estimate for full year 2013 broadcast and cable TV network ad spending (N.B. 2014 estimates are not yet available) was $74.5B (slightly below Nielsen’s $78B estimate) up just 3% YoY, suggesting that streaming ads were a full 4% of total video advertising spending and on track to top 8% for 2014. Considering the push to video from traditionally static advertising social networking sites like Facebook, which saw its CY14 ad revenues up 64.5% YoY, and Twitter, which will more than likely report ad growth well over 100% for CY14, we would not be surprised to see streaming ad spending accelerate for 2015. With online now at a meaningful share of total video advertising, this growth is a bad omen for ad-supported television.
Exh 22: US Online Advertising Revenues, 2000-2014
Exh 23: Digital Advertising by Format, 1H13 vs.1H14
This perspective is corroborated by the advertising community. A survey published by the IAB suggested that nearly two thirds of agency executives and marketers questioned expected to be increasing their spending on digital video in the next year, with less than 1% indicating a reduction (Exhibit 24). In contrast, just 39% expected to be raising their TV spending, while 14% planned a reduction. Of the respondents planning to increase spending on streaming ads, 67% expected to fund that increase, at least in part, by reducing their spending on TV. Already CPMs for high quality digital ads – targeted to specific individuals on well trafficked sites and paid for only if watched – are higher than for broadcast primetime (although Superbowl ads are higher still).
Exh 24: Advertising Executive Perspectives on Video Spending, April 2014
Driving this broader acceptance for digital video advertising are new measurement tools and metrics (including online tracking from Nielsen), and a longer track record of effectiveness. Longer term, we see linking all forms of online ads – video pre-roll, social media video, display, search, etc. – together to provide a unified approach to individual consumers, then integrating that into a broader merchandizing solution that identifies attractive customers early, delivers targeted marketing messages tied to their specific interests and actions, facilitates the buying decision both online and in store, promotes customer affinity/loyalty, and removes friction from transactions. In this context, the threat of streaming video ads for traditional TV will get worse not better.
Tell Me When it Hurts
Old habits die hard, and predicting the exact peak in US TV ad revenues is tricky, but the decelerating pattern, combined with the soaring trajectory of streaming ad sales, suggests that the peak may well be at hand. Overall, TV networks derive half of their revenues from advertising, with broadcast networks more dependent than cable nets (Exhibit 25). The media companies that own those networks typically have other businesses which reduce their overall exposure to TV ad revenues to various degrees. Ranking major media companies by advertising exposure, Scripps leads the pack with 68% exposure, followed by Discovery, CBS, Disney (not including theme parks), NBCUniversal (not including Comcast’s cable business), and Viacom, with Disney and Time Warner showing as the least exposed. Time Warner’s recent spin off of the Time Inc. publishing business further decreased its exposure to advertising. Considering analyst expectations for accelerating revenue growth across the board for media companies, it is likely fair to assume that the consensus does not include a scenario under which TV advertising declines. Despite the clear opportunities for the networks to monetize their programming through other channels, including online streaming, we believe that the weight of their advertising dependence and increasing the content costs related to new competition will be too great for media stocks to escape unscathed, particularly given their historically high relative valuations. We are also concerned for local TV station owners, who have even greater dependence on advertising revenues, but note that upcoming FCC auctions will allow them to monetize their valuable spectrum holdings, circumstances that are likely to dominate trading in their stocks.
Exh 25: Major media company exposures to advertising
The flip side of this story is the extraordinary growth of digital advertising, and in particular, the video, social and mobile ad categories (N.B. which are not mutually exclusive). We are bullish on the prospects created by shifting advertising budgets, with Google, Facebook, and Twitter all listed in our 15 stock large cap recommended model portfolio, with Twitter selected as one of our 5 top picks for 2015 (link). To a significantly lesser extent, companies like Yahoo, AOL, Pandora and IAC may also be beneficiaries of the online ad boom.