Bing And Facebook Start Down A Very Frightening Social Media Analytics Path

Finding and analyzing the collective thoughts in all the conversations happening in social media today has been a retail goal for several years now. Not coincidentally, that's exactly how long retail has failed in doing anything meaningful with that data. This week, though, an ISV and Microsoft's Bing search engine are at least making noises as though they are making a little progress.

Bing on Monday (May 16) said it is working with Facebook to use a small portion of those social site discussions—limited to the ones on Facebook and further limited to the people in the friends list of that Web searcher—to help provide more valuable results to consumers.

"The best decisions are not just fueled by facts, they require the opinions and emotions of your friends," said Yusuf Mehdi, senior vice president at Bing. "Search is now more than a fact finder. We're marrying fact-based search results with your friends' street smarts to combine the best data on the Web with the opinions of the people you trust the most and the collective IQ of the Web."

The idea of aggregating the shopping and other experiences of a closed community is a good one, with lots of potential to boost the meaningfulness of such results.

There's also a downside with this aggregation approach, namely that most consumers trust different friends to very different degrees. A single datapoint crafted from the combined actions of 50 people you absolutely trust and 50 people you personally know are dumber than a rock is likely to be no more valuable than the non-socially-aided original Bing results.

Then again, that's a mathematically correct conclusion. Will a socially fueled engine strike most consumers as more valuable, even if it isn't? And if it does strike them that way, will it make them more comfortable with purchasing whatever the groupthink recommends? It's long been said that the real enemy of sales is not a competitor's offering as much as the customer opting to make no purchase at all, often because the customer is confused or uncertain. It certainly seems plausible that this social approach from Bing could help there.

Meanwhile, a Tuesday (May 17) introduction from a software firm that touts Walgreens, Safeway, David's Bridal and as customers threatens to make even deeper inroads into social data-mining. That vendor, Attensity, said it can now search all social communications—both private, such as customers E-mailing or otherwise interacting with a retailer, and public, which is a customer posting on her Facebook page that she just received a big raise—and find new information "hidden within the unstructured text of customer conversations."

This claim is simultaneously exciting and deeply disturbing. Let's step back for a moment. For social data-mining, there are three relevant categories.For social data-mining, there are three relevant categories. One: All Aggregate. This approach, which is similar to the one Wal-Mart paid some $300 million for when it bought Kosmix last week, offers various reviews about how well Kosmix can handle massive social media data-crunching accuracy. It is the safest approach for avoiding privacy backlash, and it also gives good overall emotional reactions to products. Think of it as the world's largest focus group, with all the pluses and minuses of that research magnified by several orders of magnitude.

Two: Automatic Opt-In. This includes communications that customers and prospects engage in with your brand willingly and deliberately, such as direct E-mails, text messages, help-desk calls, tracked Web site usage, etc. A little bit of privacy pushback, but most consumers today assume this is already happening. (Ahhhh, they have far too much faith in retail corporate IT budget generosity.) It is a very accurate technique, but it's not really social data-crunching.

Three: USELT, which stands for Unlimited Scope, Extremely Limited Target. This is the nuclear-powered privacy powder-keg. It's also the ultimate in CRM. This approach is where spiders search everywhere in social media, reading every blog post and comment, Facebook notation, Twitter tweet, LinkedIn update, iTunes song selection and YouTube video diary entry. And those spiders are on a mission to find entries relating to specific existing customers and specific customer prospects.

What if a chain wanted to limit its coupons to customers who wouldn't otherwise make a purchase? What if it searched for customers who had just been laid-off, so they could be sent the coupons? Even worse, what if it sought customers who had just gotten raises or big bonuses to make sure that those customers did not get a coupon?

The idea of trying to detect a consumer's mood for pricing strategy is not new, but this type of social network data-mining could make it practical in a way that it never really was. A spokesperson for Attensity said the software absolutely can jump into the sensitive USELT arena. Should retailers go there? That's a much more complex question than it seems.

Ultimately—say, in about eight years—this won't be much of an issue. Consumers will accept it. If they want to remain anonymous, there are ways today to do so and there will be far more such options in the coming years. But near-term, that's very different. The information would be fabulous, and it could address improving margins beyond just improving revenue. If customers in 2011 even think that their favorite chain is doing this to them, the damage could be irreversible.