The day before Facebook went public, General Motors cancelled $10 million in Facebook ads. General Motors advertising execs claimed that their display ads were ineffective and failed to generate a satisfactory return-on-investment. This was the ultimate smackdown to Facebook coming on the eve on its IPO. Facebook remained silent, and did not comment. Whether GM's ad cancellation affected Facebook's IPO remains to be seen, but it triggered a firestorm of controversy and wave of discussion among social media experts and Wall Street analysts and the digital advertising community. Everybody wants to know: Are Facebook ads effective in generating sales?
If there’s one question on which much of Facebook’s $60-billion market valuation hangs (down from $104.5 billion since the IPO day), it is whether the kind of “social advertising” the social giant offers to brands actually works or not— in other words, whether having fans and social discussion around a brands product translates into actual measurable sales.
Facebook has now released some actual data from media tracking firm comScore that says it proves the value of building up a fan base on its platform, since doing so appears to increase the likelihood that a user will buy something later. But will the research convince advertisers to devote more time and money to Facebook’s social campaigns? And if so, how much of that will benefit Facebook directly?
The comScore study, which is called “The Power of Like 2: How Social Marketing Works,” is the second in a series the web-analytics firm has done with Facebook. The first report came out last July, and argued that brands using the social network need to do more than simply build up a large fan base — they need to use a combination of paid and “earned” media (that is, content that is shared voluntarily by users) to promote whatever marketing message they are focusing on. The latest report is an extension of that case, with some statistical database on what Starbucks and Target have seen from their Facebook campaigns.
Fans of a brand buy more, and so do their friends
According to comScore's new study, Starbucks saw a “statistically significant” improvement in purchasing behavior in its stores in the weeks following exposure to promotional content on Facebook. Perhaps most important of all, the analytics firm said this behavior was seen not just among those who were already fans of the brand on the social network, but also among friends of those fans — evidence of what comScore called a “latent branding impact.” The same kind of impact was seen in a study of buying behavior at Target stores, comScore said.
In a nutshell, the report says that by the fourth week following the exposure of fans and friends of fans to certain advertising content — whether in a “sponsored story” or some other social ad format — the test group's purchasing rate of 2.12 percent was a little over half a percentage point higher than the control groups’ rate. According to comScore, that means the social advertising on Facebook drove an increase in actual sales of almost 40 percent.
As Peter Kafka of All Things Digital notes, the comScore research is a double-edged sword for Facebook, since it shows that “earned media” — that is, the kind of social sharing that in many cases brands don’t even have to pay for — can generate a substantial bump in sales all by itself, without the need for traditional display ads. Theoretically, that’s the kind of ammunition brands like General Motors could use to justify dropping their ad spending on Facebook and relying on social sharing of their marketing content instead.
Facebook display ads work too, says comScore
One of the comScore study’s conclusions seems to be aimed directly at this idea — and also at critics who question whether Facebook’s paid ads are effective when the click-through rates on them are so low (even lower than the rates on generic web advertising). The report notes that an analysis of the data showed “statistically significant” increases in both online and in-store purchasing for a major retailer after exposure to display ads, despite the lack of clicks, and that
"This highlights the importance of using view-through display ad effectiveness in a medium where click-through rates are known to be lower than average.”
Facebook’s Brad Smallwood, head of measurement and insight for the social network, was more blunt in a comment to the Wall Street Journal about the results of the comScore research, saying it proved that
“It’s a myth that Facebook advertising doesn’t work.”
The Journal also noted that the quiet period following its initial stock offering has ended, so Facebook is now able to respond to some of the criticisms that arose during the IPO roadshow, and the comScore study is clearly part of that effort.
One thing the study also reinforces is just how much advertisers are betting on Facebook: according to comScore’s analysis, more than 15 percent of all U.S. online display ads were “socially enabled,” meaning they contained a message asking viewers to “like” or follow the brand or the campaign on Facebook. That’s almost double the number of ads that contained those kinds of messages in November of last year, the report said. That kind of bet is what drove Salesforce to sepend close to a billion dollars to acquire Buddy Media, which specializes in managing Facebook pages and social campaigns.
COMMENTARY: A June 5, 2012, article in Reuters titled “Facebook Comments, Ads Don't Sway Most Users: Poll,” cited the following statistic:
“Four out of five Facebook users have never bought a product or service as a result of advertising or comments on the social network site.”
In a blog post dated June 7, 2012, comSCORE'S Andrew Lipsman summarized the findings of the "Power of Like 2 report and came to the defense of Facebook over the Reuter's article by writing:
"In this particular case, it appears that the research method used was a survey, which asked users about whether or not they had ever been influenced to purchase as a result of exposure on Facebook. While surveys can be useful in assessing ad effectiveness lifts across attitudinal dimensions such as brand awareness, favorability and purchase intent, people tend not to provide very accurate assessments of their own behavior. And their accuracy in recalling their own behavior over an extended period of time can be especially unreliable. People might be able to accurately tell you how many times they have eaten at a restaurant in the past week, but they would probably do a poor job estimating that number over the past three months."
Furthermore, Lipsman disputed reports that users are spending less time on Facebook by saying,
"This inability to accurately recall past behavior also seems to be evident in another survey response where a higher percentage of Facebook users say they are spending less time on the site today vs. six months ago. comScore’s behavioral measurement of engagement, where time spent on sites is electronically and passively observed, indicates the opposite – that time spent per user is actually up a few percent in that period. In the case of the internet, people spend time doing dozens if not hundreds of things online each day. It is highly unlikely that their recall of the exact sites they visited, the amount of time they spent there or their specific exposure to brand messages will be closely aligned with what actually happened."
The audacity of comScore's Andrew Lipsman to completely disregard the accuracy of surveys because, "It is highly unlikely that their recall of the exact sites they visited, the amount of time they spent there or their specific exposure to brand messages will be closely aligned with what happened," sparks of arrogance. His blanket comment assumes that all of the hundreds or thousands of individuals surveyed for the polls about Facebook have problems recalling how much time they spent on Facebook six months ago compared to today. How does he know this to be a fact? If you were to ask me how much time I spent on Facebook six months ago, the answer would be the same: Today, I spend the vast majority of my time visiting social networks Google+ and Twitter, as I did six months ago.
The Issues With comScore Report About Facebook
I have some other problems with the new comScore "The Power of Like 2" report:
- Facebook is a client of Facebook. I would've preferred to see an independent third-party prepare a study of the effectiveness of Facebook advertising in generating sales. This would eliminate any bias.
- The timing of the report was obviously made in order to refute everything and anything bad that has been said, survyed or polled regarding Facebook.
- comScore's "The Power of Like 2" is too narrowly focused to be 100% conclusive. comScore tracked the online behaviors of the fans and friends of fans of only a few accounts: Starbucks, Target, Staples, Amazon.com, BestBuy and Skittles. It's possible that comScore "cherry-picked" these brands with the intention of producing the best results for Facebook. I am not saying that this is what occurred, but we will never know for sure.
- comScore uses the four-week data gathered for Starbucks (see above graph) as the centerpiece of the report. It's just one account out of the 30-to-50 million business accounts with a page on Facebook. Again, this highlights the weakness of relying on the findings of a report that is too narrowly focused.
- comScore uses "lift in sales" in the "The Power of Like 2" report as if it were irrefutable proof that Facebook advertising works to produce significant sales results. However, comScore did not provide any sales figures for the aforementioned brands used in the report. The report would've had more value and been more impressive if the reader had been provided actual sales figures for each brand, with and without earned media.
- comScore failed to provide how much each brand profiled in the report spent on advertising.This is the only way to measure return-on-investment. In fact, comScore skirts around the subject of ROI, preferring to emphasis "lift in sales."
- The comScore report is mostly a How-To guide on how brands can use Facebook social marketing to deliver reach, engagement and impact through a combination of paid and earned media created virally through Likes and comments. However, as a legitimate study on the effectiveness of Facebook to generate sales, I consider the resport inconclusive and inadequate for the reasons given above. There is no doubt that Facebook is a great way for firm to create brand awareness and create a large fan following. Starbucks has 25 million fans. LadyGaga even more. If your brand is successful in executing its social media marketing strategy, and you are able to successfuly create a large fan following, Facebook social marketing could work for you, since earned media does not cost you anything. However, generating earned media only works if you have a large fan following and those fans generate a lot of Likes and comments about your brand. By the way, this brings me to the next point that I am about to make regarding earned media (see below)
Why comScore's "The Power of Like 2" Could Hurt Facebook's Ad Revenues
Providing too much information can be almost as bad as not providing enough information, and comScore's "The Power of Like 2" report is a great example of providing too much. In so doing, comScore could actually end up torpedoing Facebook's advertising revenue potential, because the report is more or less a step-by-step guide of how Facebook's social advertising works and how to create "earned media." Earned media is a form of "free media" that occurs virally everytime a brand's Facebook fans click the "Like" button or make a comment about the brand. According to comScore, a brands fans or friends of fans can be influenced by these Likes and comments to make a purchase of that brands products, so much so in fact, that it creates a measureable lift in sales (using Starbucks as an example). This is a good thing to know, but not every brand has the fan following of the brands used in the report. Furthermore, by emphasizing earned media as a key contributor to create "sales lift," brands may not have an incentive to pay for advertising or could reduce their spending on ads, once they reach a critical mass necessary to create earned media virally.
Courtesy of an article dated June 12, 2012 appearing in Gigaom, an article dated June 7, 2012 appearing in the comScore Blog, an article dated June 12, 2012 appearing in the comScore Blog, an article dated April 23, 2012 appearing in Forbes, an article dated May 23, 2012 appearing in VentureBeat and an article dated May 24, 2012 appearing in Yahoo Finance
Comments