The simplest way for a site to guess whether another site is influencing purchases is to review referral link logs, to see where customers were right before. That method, however, doesn't make much sense with many social sites. A Twitter visitor, for example, will likely click on a link to read the reviews/photos/thoughts of a Twitter connection, and only after that—and perhaps one or two more links—will the customer visit the retailer's site.
The report, from an E-Commerce vendor called RichRelevance, is interesting in that its methodology appears mostly valid and it tackled questions that most others sidestep. That said, the E-Commerce influence of social media sites—especially one as narrowly focused as Twitter—has always been next to impossible to quantify.
Given that a retailer's goal is to understand what is causing its sales, limiting the analysis to direct visits is unwise. Coupling this detail with the study's conclusion that the identifiable Twitter activity was sharply higher, should demonstrate that the social value is likely being dramatically underestimated.
Let's look at the numbers. Referral percentages ranged wildly, from Google being responsible for 80.62 percent of E-Commerce traffic to Twitter leading to 0.02 percent of traffic—that's two one-hundredths of a percent. In between were Yahoo (9.67 percent), Bing (7.45 percent), AOL (1.74 percent) and Facebook (one half of one percent).
You really have to look at those social figures—especially Facebook—and wonder if they could possibly reflect reality, particularly in the teenage and youthful shopper segment. Referrals and testimonials from friends can have a huge impact on purchases.
When looking at how much those referred consumers spend—the average order value (AOV)—things get even more interesting. Here's how the dollars change when looking at those same six referrers: Google is the lowest (AOV $100.16) and Twitter is the highest ($121.33), with the in-between players of Facebook ($102.59), Bing ($104.62), Yahoo ($105.13) and AOL ($105.27) showing just about zero correlation between "who sends the most" and "who brings in the most money."Making this situation much more problematic is the absence of a reliable means for determining true influence. Asking people how they came to be interested in the product they are buying is fairly useless.
Most won't bother to answer truthfully, and many more won't actually know. Their kneejerk response might be, "I liked the colors/price/functionality," which is almost certainly true. But they won't add, "What got this on my radar at all was when my friend wrote about it on her Facebook page, which I found because she also tweeted about it."
One reliable method of determining influence is to create very targeted campaigns on, let's say, Twitter and see who shortly shows interest in it. But that marketing approach simply misses the point of social media. The influence referenced above in the kneejerk response example was exclusively based on the relationship the customer had with his/her friend. It's a personal referral sale. Had it been a tweet or a Facebook post from a retailer or manufacturer, the influence would likely be far less. Regrettably, this stuff has to happen organically for it to have the huge impact on sales that it has.
RichRelevance's chief marketing officer, Diane Kegley, concedes the challenge based on her own company's report.
"It seems like some merchants have Twitter as part of their marketing value chain, but quantifying the return on that specific kind of behavior is problematic," she said in an E-mail. "More generally, it highlights the need for merchants to correctly understand attribution. While we would like consumer behavior to be the result of simple cause-effect chains, we know—in no small measure from our own personalization algorithms—that it is, in fact, much more complex."
Complex is just one small part of it. "Complex" suggests that a sophisticated enough algorithm could crunch the data and expose all of the social influence, while "impossible to quantify precisely" is probably closer to the sad truth. But the more solid part of this report—confirming that social site buyers (on those rare occasions when they can be identified)—makes clear that this problem must get figured out. The potential dollars here are huge.
"When we factor shopper intent into the equation, the fact that Twitter users generate higher [purchase values] makes sense. Twitter users—similar to Facebook users—are not generally researching or completing purchases. A Twitter user, however, might see a tweet from a friend or an ad from a retailer and get pulled into a retailer's Web site. This spontaneity, combined with influence from word of mouth or an ad, can lead to a higher average order value."Kegley also argues that it's not merely buying more items or spending more in total for that purchase. The power of the referral is causing consumers to be less focused on getting the best price. In short, they're willing to pay more when coming from a Twitter or Facebook referral than when coming from other sources. (Do we have your attention now, you margin-obsessed product line managers?)
"The fact that Twitter users may be going online to share information with their social networks—not with intent to purchase—may explain the decreased frequency of shopping," Kegley said. "But when they are inspired to purchase, spontaneity may play a role in decreased price sensitivity."
There actually is a way to track social purchases, but it's hard. It's analyzing the millions of references to your products in social media and simply correlating that information with the purchases. Indeed, it's the exact type of analysis we're expecting to see from Wal-Mart and its April $300 million Kosmix acquisition.
That's just the first step, though. It merely tells me the influences behind these sales, which we can already take a really good guess at. But doing something about this to hike sales even more—to influence the influencers, if you will—can be dangerous.
There's the huge risk of killing the goose that laid the golden AOV. Any visible attempt to change or incentive any conduct on social media could have the disastrous impact of lessening that influence. How would you feel if a friend was praising some new HP service and you started seriously considering it—on the basis of your friend's knowledge—and you then discovered that, although she never said anything about it, she was getting a $1,000 every time a friend of hers bought one? Would it sour that service in your mind, even though none of the facts about the service has changed? Would it decrease—nay, obliterate—any credibility you attribute to any business post from that friend forever more?
The only effective approach for businesses is to watch closely and be ready to act on organic social activity. For example, if your systems detect some strong endorsements of some new products/services, think about on-the-fly pricing changes. Do you drop prices to turn those interests into lots of quick sales? Do you increase the price, factoring in the weaker price sensitivity of people coming in from strong social referrals? It might be an early heads-up. But social sales are likely to happen too quickly—and far too unreliably—for any chain to be able to make meaningful use of that information.
Social media is a monster of an influencer, and the monster part is that it's both huge and will punish you for any visible attempt to take advantage of it. Whoever said business life was supposed to be fair?