Ad Battle of the Century: Facebook crushes Primetime TV

The social media era began in 2006, when Facebook first welcomed everyone over 13 and Twitter launched. We’ve been waiting six years now for some useful way of comparing the efficiency of traditional ads versus digital advertising. Such a method is now in sight.

When I say “advertising,” of course, I’m not talking solely about what marketers almost always mean by the word “advertising:” paid ad units, whether on TV, in print, on Facebook or anywhere else on the web. Paid ad forms have been measured and compared to death. But it is virtually impossible to find any study of the advertising impact of non-paid messages.

Focusing exclusively on paid ads is not really relevant any longer. Increasingly, advertising messages are being delivered over digital media in non-advertising forms—brand-owned content spread across social platforms by paid and earned media. These messages may be blog stories, Facebook posts, YouTube films and so on. They may be created by brands or by ordinary folks who like a particular brand. Research shows that many of these non-ad forms are far more persuasive and effective than ad units. (I’ll get to that research in a minute.)

So we need an approach to measuring non-ad units in ways that make them comparable to traditional advertising. This is critical because the rapidly changing present and future of advertising is all about assembling large audiences around valuable brand-owned content that delivers powerful brand stories across platforms—stories so compelling that people share them with their social networks. Let’s call it the “new advertising.” Let’s call it “socially multiplied advertising.” A Minneapolis agency calls it “non-advertising.” My agency has been calling it “post-advertising” for the past 7 years. Whatever we call it, let’s face it: Eventually we’re going to call it just plain “advertising” because this is what advertising is becoming.

Fundamentally, we’re looking at the use of paid, owned and earned media to create large communities of people who will share brand recommendations and strongly influence one another’s behavior.

Redefining advertising this way is really the only “new” thing in this story. The rest is arithmetic. So I want to walk through a set of calculations and the research supporting these calculations. Together they establish a startling (to me, at least) conclusion: Socially multiplied advertising, the core of which is contagious brand-owned content on social platforms, is roughly 100 times more cost-efficient for brand advertising than primetime TV on a cost-per-thousand impressions (CPM) basis.

I concede readily that this is an initial effort and aims solely at arriving at an approach to measuring social advertising and an order-of-magnitude approximation that agencies and clients can study and refine. So argue with this, please. But make your arguments fact- and research-based. No emotional inveighing about the “power of television.” If everyone at the networks thought people loved TV spots, they wouldn’t all be suing Dish Network to stop its new ad-skipping technology.

Only recently has third-party research emerged to support a numbers-based comparison of social advertising versus television with enough accuracy to be useful. One key piece of recent research that I point to often is the Nielsen Global Trust in Advertising survey, begun in 2007. Earlier this year, the survey’s 2012 edition revealed that 92% of consumers worldwide trust recommendations of friends and family when making purchase decisions and 70% trust recommendations from strangers online. The same survey showed that, at most, only 47% of consumers trust traditional ads—TV, newspaper and magazines. Trust in such traditional ads has dropped off a cliff in the past three years, according to Nielsen.

Nielsen’s data suggests that having a friend or stranger deliver a brand message online through social media is 1.5 to 2 times more likely to influence behavior than delivering the same message in a TV ad. (So if you were thinking TV is more impactful or more persuasive than a post on Facebook, you’re just wrong, according to Nielsen.)

The scholarly journal Nature in September published a research paper by social scientists from the University of California and Facebook’s Data Science department (“A 61-million-person experiment in social influence and political mobilization”) reporting on their study of a single message delivered on Facebook to encourage voting in the US 2010 off-year elections. In addition to finding that the social message materially influenced voting, and that sharing spreads through multiple layers of friends of friends, the study concludes that pass-along recommendations of “close friends exerted about four times more influence on the total number of…voters mobilized than the message itself.”

As Fordham University’s Philip Napoli, a leading media theorist and researcher, pointed out to me recently, this is consistent with roughly eight decades of media theory and research about how conversations and mass media interact to influence people’s behavior (Google “two-step flow”). More importantly, it confirms the outsized influence exerted when people share non-advertising or post-advertising forms of advertising. In other words, all the research is saying a brand will get far more impact if a friend or a stranger delivers the brand’s message on social media than if the brand delivers the message directly. This works whether the message is about voting or corn flakes.

Facebook itself is just beginning to release data that will allow marketers to calculate how many brand impressions are delivered when someone “likes” a brand page or shares a post.

Looking at the basic data, each Facebook user worldwide has 190 friends on average, according to a paper called Anatomy of Facebook, published by the social network itself a year ago. This may be low. Pew found this spring that the average number of friends is 229 in the US. There’s data indicating it’s even higher in the UK.

More importantly, Facebook recently said that 16% of a user’s Facebook friends actually see a post from a brand (or one shared or liked by a friend) on the brand’s or friend’s page or in their own newsfeed or ticker. Theoretically, Facebook’s own experts tell us, 100% of a brand’s fans would see all posts if they had uncrowded newsfeeds. But, based on a user’s connections, settings and preferences, Facebook says its algorithms prioritize the messages delivered to the user to keep the newsfeed and ticker tidy. (This piece is already long enough, so, without taking sides, I’m going to skip the heated blogosphere debate over whether Facebook is purposefully shutting off the content spigot purely to force businesses to pay more to reach their fans.)

Even at 16%, the “virality” or pass-along percentage is significantly higher than any estimate I’d seen prior to the release of the real numbers. (Bear in mind, that 16% can be multiplied 3 to 5 times by paying Facebook to make the post a sponsored story.)

Facebook, however, says it has no data on the way messages are shared among friends of friends. (I call that the third-level or tertiary audience. Fans are primary; fans’ friends are secondary; friends of those friends are tertiary.)

Using Facebook data, independent research, an educated guess on the drop-off of third-level sharing and our own extensive experience with social media programming, Story recently created a model to compare the delivery of effective impressions on a TV campaign versus a Facebook campaign.

Our model shows brand impressions on US prime time network TV cost about 180 times what the same impressions cost in a social media campaign. In the UK, using ITV’s “Coronation Street” as the baseline, TV CPMs are roughly 70 times more expensive than social media costs per thousand impressions.

The driving force behind this yawning gap, of course, is the extent to which social sharing spreads messages at low incremental costs to advertisers. A well run, year-long Facebook campaign aimed at 1 million fans can generate 4 to 5 billion impressions while a similarly well run TV campaign delivers less than 100 million. This is the power of social multipliers.

Importantly, our model’s calculations are consistent with one set of actual metrics that Facebook provided recently to Story Media, our planning and buying arm, for a US client—national cable station WGN America, which advertises to increase viewership among its target audiences.

Here’s what happened with WGNA over the 12 months ending in August 2012, according to Facebook’s metrics: We used paid ads and promoted and unpromoted Facebook posts to build a community of 1.8 million followers. Daily, we posted contagious content that maintained high engagement scores from the brand’s fans. Our efforts, measured by standard analytical tools, generated 4.3 billion impressions at a cost of 15 pence per thousand impressions. The result was exactly what the client wanted—significantly higher Nielsen ratings and a marked demographic shift to a younger audience.

That 15p CPM compares to a £14.97 CPM on UK television (103 times more expensive) and £38.09 on primetime in the States (262 times more expensive).

The model we created to project the results of a social campaign includes fans, their friends and the friends’ friends. It produced an estimate of total impressions 15% higher than the measured results—4.9 billion. At a CPM of 21p, the cost projected in the model is consistent with the actual case. The main difference between the model and the actual cost numbers is that we assumed higher costs than we actually encountered. (On average, the size of the communities in the two cases was very close.)

All these calculations factor in content costs, reasonable or actual production costs, the relative trust (or influence) of the different media (as measured by Nielsen) and the pass-along impressions gained through earned media. I know knowledgeable, reasonable people may differ over the precise values that ought to be assigned to each of these factors in a variety of different cases. But even if our calculations are off by 50%—and after conferring with several academic experts I am convinced they are not—the advantages of social media in delivering brand impressions is inescapable and staggering.

The model shows that the key variables in producing results from social media are the extent of third-level sharing and the quality of the content being posted by the brand.

In the Facebook examples, the most important controllable factor in the model is the “people are talking about this” or PTAT score, which Facebook says is a measurement of content quality and is the best engagement score for their pages. (PTAT measures likes, shares, commenting and other so-called “high-value tasks” that show engagement with a piece of content. Importantly, it reports engagement by giving the hard number of “people talking about this” — not a percentage. )

Tellingly, at when the PTAT score is 1% to 2% of the total community, fan communities of brand pages grow organically either not at all or very little. Unfortunately for brand marketers, most brand pages score this low or lower because the content is poor or not refreshed daily or, usually, both, so the audience doesn’t interact with the page.

But at a 10% PTAT or better, organic growth in audience size and the velocity of sharing both follow a steep upward curve, generating literally billions of extremely low-cost, high-impact impressions from a community of a million or more Facebook fans. The Facebook brand pages which Story programs for clients generally hit this mark, varying from roughly 9% to 12% PTAT and higher. WGNA, our client in the example I cited above, has enjoyed PTAT scores of 12.5% recently. Those scores have been as high as 20% or more in the past.

We’re going to keep working to refine the art of modeling impact, results, and CPMs in socially multiplied advertising. We especially want to drill down beyond the friends of fans to get real data on the extent of sharing deeper in the network.

We’re not predicting the end of TV, by the way. (We love TV!) But we do know that a radical change in the marketing media mix is coming, along with a shift from old advertising to post-advertising.

Experts have said repeatedly chided CMOs for clinging to the past, arguing that marketers need to overcome long-established habits and embrace the smart use of non-traditional forms, either in combination with traditional advertising or not Such experts include Rex Briggs in his new book SIRFs Up: Catching the Next Wave of Marketing and independent researchers in a Journal of Advertising Research paper titled “The Power of Inertia.”

These necessary and important changes in approach and spending will be driven by a deep understanding of the comparative numbers that matter—costs and results. So my modest proposal is this: Let’s start doing the numbers.

We intend to work with academics, advertising practitioners and, we hope, Facebook’s own data experts. We’d love for others to join the effort. If you’re interested in participating, have questions or want to contribute any ideas or critiques of our approach, just shoot me an email at